michael@0: michael@0: CHANGES BETWEEN 2.5.2 and 2.5.3 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - A vulnerability was identified and fixed in the new CFF driver michael@0: (cf. http://savannah.nongnu.org/bugs/?41697; it doesn't have a michael@0: CVE number yet). All users should upgrade. michael@0: michael@0: - More bug fixes related to correct positioning of composite michael@0: glyphs. michael@0: michael@0: - Many fixes to better protect against malformed input. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - FreeType can now use the HarfBuzz library to greatly improve the michael@0: auto-hinting of fonts that use OpenType features: Many glyphs michael@0: that are part of such features but don't have cmap entries are michael@0: now handled properly, for example small caps or superscripts. michael@0: Define the configuration macro FT_CONFIG_OPTION_USE_HARFBUZZ to michael@0: activate HarfBuzz support. michael@0: michael@0: You need HarfBuzz version 0.9.19 or newer. michael@0: michael@0: Note that HarfBuzz depends on FreeType; this currently causes a michael@0: chicken-and-egg problem that can be solved as follows in case michael@0: HarfBuzz is not yet installed on your system. michael@0: michael@0: 1. Compile and install FreeType without the configuration michael@0: macro FT_CONFIG_OPTION_USE_HARFBUZZ. michael@0: michael@0: 2. Compile and install HarfBuzz. michael@0: michael@0: 3. Define macro FT_CONFIG_OPTION_USE_HARFBUZZ, then compile michael@0: and install FreeType again. michael@0: michael@0: With FreeType's `configure' script the procedure boils down to michael@0: configure, build, and install Freetype, then configure, compile, michael@0: and install HarfBuzz, then configure, compile, and install michael@0: FreeType again (after executing `make distclean'). michael@0: michael@0: - All libraries FreeType depends on are now checked using the michael@0: `pkg-config' configuration files first, followed by alternative michael@0: methods. michael@0: michael@0: - The new value `auto' for the various `--with-XXX' library michael@0: options (for example `--with-harfbuzz=auto') makes the michael@0: `configure' script automatically link to the libraries it finds. michael@0: This is now the default. michael@0: michael@0: - In case FreeType's `configure' script can't find a library, you michael@0: can pass environment variables to circumvent pkg-config, and michael@0: those variables have been harmonized as a consequence of the michael@0: changes mentioned above: michael@0: michael@0: LIBZ -> removed; use LIBZ_CFLAGS and LIBZ_LIBS michael@0: LIBBZ2 -> removed; use BZIP2_CFLAGS and BZIP2_LIBS michael@0: LIBPNG_LDFLAGS -> LIBPNG_LIBS michael@0: michael@0: `./configure --help' shows all available environment variables. michael@0: michael@0: - The `freetype-config' script now understands option `--static' michael@0: to emit static linking information. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.5.1 and 2.5.2 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Improving the display of some broken TrueType fonts introduced a michael@0: bug that made FreeType crash on some popular (but not fully michael@0: conformant) fonts like `ahronbd.ttf'. michael@0: michael@0: - Another round of improvements to correct positioning and hinting michael@0: of composite glyphs in TrueType fonts. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - Version 2.5.1 introduced a bug in handling embedded bitmap michael@0: strikes of TrueType fonts, causing garbage display under some michael@0: circumstances. michael@0: michael@0: - The `ftgrid' demo program couldn't be compiled in michael@0: non-development builds. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.5 and 2.5.1 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - For some WinFNT files, the last glyph wasn't displayed but michael@0: incorrectly marked as invalid. michael@0: michael@0: - The vertical size of glyphs was incorrectly set after a call to michael@0: `FT_GlyphSlot_Embolden', resulting in clipped glyphs. michael@0: michael@0: - Many fields of the `PCLT' table in SFNT based fonts (if accessed michael@0: with `FT_Get_Sfnt_Table') were computed incorrectly. michael@0: michael@0: - In TrueType fonts, hinting of composite glyphs could sometimes michael@0: deliver incorrect positions of components or even distorted michael@0: shapes. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - WOFF font format support has been added. michael@0: michael@0: - The auto-hinter now supports Hebrew. Greek and Cyrillic support michael@0: has been improved. michael@0: michael@0: - Support for the forthcoming `OS/2' SFNT table version 5, as can michael@0: be found e.g. in the `Sitka' font family for Windows 8.1. michael@0: michael@0: - The header file layout has been changed. After installation, michael@0: all files are now located in `/include/freetype2'. michael@0: michael@0: Applications that use (a) `freetype-config' or FreeType's michael@0: `pkg-config' file to get the include directory for the compiler, michael@0: and (b) the documented way for header inclusion like michael@0: michael@0: #include michael@0: #include FT_FREETYPE_H michael@0: ... michael@0: michael@0: don't need any change to the source code. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - The stem darkening feature of the new CFF engine can now be michael@0: fine-tuned with the new `darkening-parameters' property. michael@0: michael@0: - `ftgrid' has been updated to toggle various engines with the `H' michael@0: key, similar to `ftview' and `ftdiff'. michael@0: michael@0: - The functionality of `ttdebug' has been greatly enhanced. michael@0: michael@0: . It now displays twilight, storage, and control value data; key michael@0: `T' shows the twilight point table, key `S' the storage data, michael@0: and key `C' the control value table. michael@0: michael@0: . Some keys have been reassigned from lowercase to their michael@0: uppercase equivalents; for example `q' to quit the program is michael@0: now `Q'. michael@0: michael@0: . Key `f' finishes the current function. michael@0: michael@0: . Key `R' restarts the debugger. michael@0: michael@0: . Keys `b' and `p' set a breakpoint. michael@0: michael@0: . Key `B' provides a function call backtrace. michael@0: michael@0: - Better support of ARMv7 and x86_64 processors. michael@0: michael@0: - Apple's `sbix' color bitmap format is now supported. michael@0: michael@0: - Improved auto-hinter rendering for many TrueType fonts, michael@0: especially in the range 20-40ppem. michael@0: michael@0: - A new face flag `FT_FACE_FLAG_COLOR' has been added (to be michael@0: accessed with the macro `FT_HAS_COLOR'). michael@0: michael@0: - `FT_Gzip_Uncompress' (modeled after zlib's `uncompress' michael@0: function) has been added; this is a by-product of the newly michael@0: added WOFF support. michael@0: michael@0: - Support for a build with `cmake' has been contributed by John michael@0: Cary . michael@0: michael@0: - Support for x64 builds with Visual C++ has been contributed by michael@0: Kenneth Miller michael@0: michael@0: - Manual pages for most demo programs have been added. michael@0: michael@0: - The GETINFO bytecode instruction for TrueType fonts was buggy if michael@0: used to retrieve subpixel hinting information. It was necessary michael@0: to set selector bit 6 to get results for selector bits 7-10, michael@0: which is wrong. michael@0: michael@0: - Improved computation of emulated vertical metrics for TrueType michael@0: fonts. michael@0: michael@0: - Fixed horizontal start-up position of vertical phantom points in michael@0: TrueType bytecode. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.4.12 and 2.5 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - The cache manager function `FTC_Manager_Reset' didn't flush the michael@0: cache. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - Behdad Esfahbod (on behalf of Google) contributed support for michael@0: color embedded bitmaps (eg. color emoji). michael@0: michael@0: A new load flag, FT_LOAD_COLOR, makes FreeType load color michael@0: embedded-bitmaps, following this draft specification michael@0: michael@0: https://color-emoji.googlecode.com/git/specification/v1.html michael@0: michael@0: which defines two new SFNT tables, `CBDT' and `CBLC' (named and michael@0: modeled after `EBDT' and `EBLC', respectively). The color michael@0: bitmaps are stored in the new FT_PIXEL_MODE_BGRA format to michael@0: represent BGRA pre-multiplied sRGB images. If PNG support is michael@0: available, PNG color images as defined in the same proposed michael@0: specification are supported also. michael@0: michael@0: Note that color bitmaps are converted to grayscale if client michael@0: didn't ask for color. michael@0: michael@0: - As announced in the previous release, the old FreeType CFF michael@0: engine is now disabled by default. It can be conditionally michael@0: compiled by defining the configuration macro michael@0: CFF_CONFIG_OPTION_OLD_ENGINE. michael@0: michael@0: - As announced in the previous release, all code related to macro michael@0: FT_CONFIG_OPTION_OLD_INTERNALS has been removed, thus becoming michael@0: obsolete. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - The property API (`FT_Property_Get' and `FT_Property_Set') is michael@0: now declared as stable. michael@0: michael@0: The exception, however, are the experimental auto-hinter michael@0: properties `glyph-to-script-map' and `fallback-script' which are michael@0: subject to change in a forthcoming release. michael@0: michael@0: - `ftview' has been updated to support color embedded bitmaps; it michael@0: can be toggled on and off with key `c'. The small cache toggle michael@0: is now key `K'. michael@0: michael@0: - It is now possible to control the version of the TrueType michael@0: hinting engine using the new `interpreter-version' property of michael@0: the `truetype' module: Versions 35 and 38 (the default) are michael@0: supported, which roughly corresponds to disable and enable michael@0: subpixel hinting support, respectively. michael@0: michael@0: In both `ftview' and `ftdiff', switching between the two michael@0: versions can be done with key `H'. In the `ftbench' demo michael@0: program, command line option `-H' has been extended to activate michael@0: the non-default interpreter version. michael@0: michael@0: - The `ttdebug' program has been further improved. In particular, michael@0: it accepts a new command line option `-H' to select the hinting michael@0: engine. michael@0: michael@0: - Another round of TrueType subpixel hinting fixes. michael@0: michael@0: - The `apinames' tool can now create an import file for NetWare. michael@0: michael@0: - 64bit compilation of the new CFF engine was buggy. michael@0: michael@0: - Some fixes to improve robustness in memory-tight situations. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.4.11 and 2.4.12 michael@0: michael@0: - We have another CFF parsing and hinting engine! Written by Dave michael@0: Arnold , this work has been contributed by michael@0: Adobe in collaboration with Google. It is vastly superior to michael@0: the old CFF engine, and it will replace it in the next release. michael@0: Right now, it is still off by default, and you have to michael@0: explicitly select it using the new `hinting-engine' property of michael@0: the cff driver: michael@0: michael@0: ... michael@0: #include FT_MODULE_H michael@0: #include FT_CFF_DRIVER_H michael@0: michael@0: FT_Library library; michael@0: int engine = FT_CFF_HINTING_ADOBE; michael@0: michael@0: michael@0: ... michael@0: FT_Property_Set( library, "cff", "hinting-engine", &engine ); michael@0: michael@0: The code has a (mature) beta status; we encourage all users to michael@0: test it and report any problems. michael@0: michael@0: In case you want to activate the new CFF engine unconditionally, michael@0: apply this patch: michael@0: michael@0: --- snip --- michael@0: diff --git a/src/cff/cffobjs.c b/src/cff/cffobjs.c michael@0: index ebcf189..3f2ce6b 100644 michael@0: --- a/src/cff/cffobjs.c michael@0: +++ b/src/cff/cffobjs.c michael@0: @@ -1056,7 +1056,7 @@ michael@0: michael@0: michael@0: /* set default property values */ michael@0: - driver->hinting_engine = FT_CFF_HINTING_FREETYPE; michael@0: + driver->hinting_engine = FT_CFF_HINTING_ADOBE; michael@0: driver->no_stem_darkening = FALSE; michael@0: michael@0: return FT_Err_Ok; michael@0: --- snip --- michael@0: michael@0: - The macro FT_CONFIG_OPTION_OLD_INTERNALS is no longer set by michael@0: default. In the next release, we will completely remove the michael@0: associated code. Please update your programs in case you are michael@0: still using this macro. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - The (top-level) `configure' script now respects the MAKE michael@0: environment variable to specify a `make' binary. For backwards michael@0: compatibility, GNUMAKE still overrides MAKE, though. michael@0: michael@0: - The `ftview' and `ftdiff' demo programs have been redesigned, michael@0: showing more options permanently on the screen, among other michael@0: minor improvements. michael@0: michael@0: - Using the `H' key, it is now possible to select the CFF engine michael@0: in both `ftview' and `ftdiff'. michael@0: michael@0: - The new command line option `-H' for `ftbench' selects the Adobe michael@0: CFF engine. michael@0: michael@0: - It is now possible to directly select the LCD rendering mode michael@0: with the keys `A'-`F' in `ftview'. The key mapping for cycling michael@0: through LCD modes has been changed from `K' and `L' to `k' and michael@0: `l', and toggling custom LCD filtering is no longer mapped to michael@0: key `F' but to key `L'. michael@0: michael@0: - In `ftdiff', key `x' toggles between layout modes: Either use michael@0: the advance width (this is new and now the default) or the michael@0: bounding box information to determine line breaks. michael@0: michael@0: - For all demo tools, the new command line option `-v' shows the michael@0: version. michael@0: michael@0: - For the demo tools with a GUI, the new command line options `-w' michael@0: and `-h' select the width and the height of the output window, michael@0: respectively. michael@0: michael@0: - The `ttdebug' program was broken and has been reactivated. Note michael@0: that this program is not compiled by default. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.4.10 and 2.4.11 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Some vulnerabilities in the BDF implementation have been fixed. michael@0: Users of this font format should upgrade. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - Subpixel hinting support has been contributed by Infinality, michael@0: based on Greg Hitchcock's whitepaper at michael@0: michael@0: http://www.microsoft.com/typography/cleartype/truetypecleartype.aspx michael@0: michael@0: Originally, it was a separate patch available from michael@0: michael@0: http://www.infinality.net/blog/ michael@0: michael@0: and which has been integrated. michael@0: michael@0: Note that ClearType support is not completely implemented! In michael@0: particular, full support for the options `compatible_widths', michael@0: `symmetrical_smoothing, and `bgr' (via the GETINFO bytecode michael@0: instruction) is missing. michael@0: michael@0: Activation of subpixel hinting support can be controlled with michael@0: the `TT_CONFIG_OPTION_SUBPIXEL_HINTING' configuration option; it michael@0: is switched off by default. This feature is still experimental; michael@0: we welcome test reports! michael@0: michael@0: - Support for OpenType collections (OTC) has been added. michael@0: michael@0: - Pure CFF fonts within an SFNT wrapper are now supported. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - Minor rendering improvements to the auto-hinter. michael@0: michael@0: - `FT_GlyphSlot_Oblique' now uses a shear angle of 12°. michael@0: michael@0: - Experimental support to handle `property modules', for example michael@0: to control the behaviour of the auto-hinter. The API consists michael@0: of two new functions, `FT_Property_Set' and `FT_Property_Get'. michael@0: michael@0: The code is still subject to change and should not be used for michael@0: production. michael@0: michael@0: - The `ftdiff' demo program now supports UTF-8 encoded input files michael@0: for option `-f'. michael@0: michael@0: - Using keys `r' and `R', you can now adjust the stroker radius in michael@0: the `ftview' demo program. michael@0: michael@0: - Other, minor fixes and improvements. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.4.9 and 2.4.10 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Incremental glyph loading as needed by ghostscript was broken. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - A new function `FT_Outline_EmboldenXY', contributed by Alexei michael@0: Podtelezhnikov. michael@0: michael@0: - In the `ftview' demo program, key `e' has been replaced with `x' michael@0: and `y' to embolden in the horizontal and vertical direction, michael@0: respectively. michael@0: michael@0: - The glyph spacing computation in `FT_GlyphSlot_Embolden' (and michael@0: similar code in `ftview') has been improved. michael@0: michael@0: - Minor improvements to the TrueType bytecode interpreter and michael@0: glyph loader, the auto-hinter, and the B/W rasterizer. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.4.8 and 2.4.9 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Another round of fixes to better handle invalid fonts. Many of michael@0: them are vulnerabilities (see CVE-2012-1126 up to CVE-2012-1144 michael@0: and SA48320) so all users should upgrade. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - The `ENCODING -1 ' format of BDF fonts is now supported. michael@0: michael@0: - For BDF fonts, support for the whole Unicode encoding range has michael@0: been added. michael@0: michael@0: - Better TTF support for x_ppem != y_ppem. michael@0: michael@0: - `FT_Get_Advances' sometimes returned bogus values. michael@0: michael@0: - The demo programs no longer recognize and handle default michael@0: suffixes; you now have to always specify the complete font name. michael@0: michael@0: - Better rendering and LCD mode cycling added to `ftview'. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.4.7 and 2.4.8 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Some vulnerabilities in handling CID-keyed PostScript fonts have michael@0: been fixed; see CVE-2011-3439. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - Chris Liddell contributed a new API, `FT_Get_PS_Font_Value', to michael@0: retrieve most of the dictionary keys in Type 1 fonts. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.4.6 and 2.4.7 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Some vulnerabilities in handling Type 1 fonts have been fixed; michael@0: see CVE-2011-3256. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - FreeType now properly handles ZapfDingbats glyph names while michael@0: constructing a Unicode character map (for fonts which don't have michael@0: one). michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.4.5 and 2.4.6 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - For TrueType based fonts, the ascender and descender values were michael@0: incorrect sometimes (off by a pixel if the ppem value was not a michael@0: multiple of 5). Depending on the use you might now experience michael@0: a different layout; the change should result in better, more michael@0: consistent line spacing. michael@0: michael@0: - Fix CVE-2011-0226 which causes a vulnerability while handling michael@0: Type 1 fonts. michael@0: michael@0: - BDF fonts containing glyphs with negative values for ENCODING michael@0: were incorrectly rejected. This bug has been introduced in michael@0: FreeType version 2.2.0. michael@0: michael@0: - David Bevan contributed a major revision of the FreeType stroker michael@0: code: michael@0: michael@0: . The behaviour of FT_STROKER_LINEJOIN_BEVEL has been corrected. michael@0: michael@0: . A new line join style, FT_STROKER_LINEJOIN_MITER_FIXED, has michael@0: been introduced to support PostScript and PDF miter joins. michael@0: michael@0: . FT_STROKER_LINEJOIN_MITER_VARIABLE has been introduced as an michael@0: alias for FT_STROKER_LINEJOIN_MITER. michael@0: michael@0: . Various stroking glitches has been fixed. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - SFNT bitmap fonts which contain an outline glyph for `.notdef' michael@0: only no longer set the FT_FACE_FLAG_SCALABLE flag. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.4.4 and 2.4.5 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - A rendering regression for second-order Bézier curves has been michael@0: fixed, introduced in 2.4.3. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - If autohinting is not explicitly disabled, FreeType now uses michael@0: the autohinter if a TrueType based font doesn't contain native michael@0: hints. michael@0: michael@0: - The load flag FT_LOAD_IGNORE_GLOBAL_ADVANCE_WIDTH has been made michael@0: redundant and is simply ignored; this means that FreeType now michael@0: ignores the global advance width value in TrueType fonts. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - `FT_Sfnt_Table_Info' can now return the number of SFNT tables of michael@0: a font. michael@0: michael@0: - Support for PCF files compressed with bzip2 has been contributed michael@0: by Joel Klinghed. To make this work, the OS must provide a michael@0: bzip2 library. michael@0: michael@0: - Bradley Grainger contributed project and solution files in michael@0: Visual Studio 2010 format. michael@0: michael@0: - Again some fixes to better handle broken fonts. michael@0: michael@0: - Some improvements to the B/W rasterizer. michael@0: michael@0: - Fixes to the cache module to improve robustness. michael@0: michael@0: - Just Fill Bugs contributed (experimental) code to compute blue michael@0: zones for CJK Ideographs, improving the alignment of horizontal michael@0: stems at the top or bottom edges. michael@0: michael@0: - The `ftgrid' demo program can now display autohinter segments, michael@0: to be toggled on and off with key `s'. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.4.3 and 2.4.4 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - UVS support (TrueType/OpenType cmap format 14) support is fixed. michael@0: This regression has been introduced in version 2.4.0. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - Detect tricky fonts (e.g. MingLiU) by the lengths and checksums michael@0: of Type42-persistent subtables (`cvt ', `fpgm', and `prep') when michael@0: a TrueType font without family name is given. The previous fix, michael@0: introduced in 2.4.3, was too rigorous, causing many subsetted michael@0: fonts (mainly from PDF files) displayed badly because FreeType michael@0: forced rendering with the TrueType bytecode engine instead of michael@0: the autohinter. michael@0: michael@0: - Better support for 64bit platforms. michael@0: michael@0: - More fixes to improve handling of broken fonts. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.4.2 and 2.4.3 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Fix rendering of certain cubic, S-shaped arcs. This regression michael@0: has been introduced in version 2.4.0. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - To fix the above mentioned rendering issue, a new spline michael@0: flattening algorithm has been introduced which speeds up both michael@0: conic and cubic arcs. michael@0: michael@0: - Handling of broken fonts has been further improved. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.4.1 and 2.4.2 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - A stack overflow in CFF Type2 CharStrings interpreter is fixed. michael@0: michael@0: - Handling Type 42 font deallocation was broken; additionally, the michael@0: library is now more robust against malformed Type 42 fonts. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - Two new functions, `FT_Reference_Library' (in FT_MODULE_H) and michael@0: `FT_Reference_Face' (in FT_FREETYPE_H), have been added to michael@0: simplify life-cycle management. A counter gets initialized to 1 michael@0: at the time an FT_Library (or FT_Face) structure is created. michael@0: The two new functions increment the respective counter. michael@0: `FT_Done_Library' and `FT_Done_Face' then only destroy a library michael@0: or face if the counter is 1, otherwise they simply decrement the michael@0: counter. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.4.0 and 2.4.1 michael@0: michael@0: I. IMPORTANT CHANGES michael@0: michael@0: - A serious bug in the CFF font module prevented display of many michael@0: glyphs in CFF fonts like `MinionPro-Regular.otf'. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.3.12 and 2.4.0 michael@0: michael@0: I. IMPORTANT CHANGES michael@0: michael@0: - Since May 2010, all patents regarding the TrueType bytecode michael@0: interpreter have expired worldwide. Consequently, we now define michael@0: TT_CONFIG_OPTION_BYTECODE_INTERPRETER by default (and undefine michael@0: TT_CONFIG_OPTION_UNPATENTED_HINTING). michael@0: michael@0: - A new function `FT_Library_SetLcdFilterWeights' is available to michael@0: adjust the filter weights set by `FT_Library_SetLcdFilter'. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - Thanks to many reports from Robert Święcki, FreeType's stability michael@0: in handling broken or damaged fonts is much improved. michael@0: michael@0: - Support for LCD filter control has been added to the demo michael@0: programs `ftdiff' and `ftview'. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.3.11 and 2.3.12 michael@0: michael@0: I. IMPORTANT CHANGES michael@0: michael@0: - For `FT_Open_Face', new parameters are available to ignore michael@0: preferred family names: FT_PARAM_TAG_IGNORE_PREFERRED_FAMILY and michael@0: FT_PARAM_TAG_IGNORE_PREFERRED_SUBFAMILY. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - Support for incremental font loading (controlled with the michael@0: FT_CONFIG_OPTION_INCREMENTAL macro) is now active by default. michael@0: michael@0: - Better support for vertical metrics. michael@0: michael@0: - Various minor bug fixes. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.3.10 and 2.3.11 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Version 2.3.10 broke PCF support. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.3.10 and 2.3.9 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - If all ASCII digits in a font have the same (unscaled) width, michael@0: the autohinter respects this and won't change it. michael@0: michael@0: - TrueType fonts are now rasterized correctly if the horizontal michael@0: and vertical resolution differ. michael@0: michael@0: - Type 1 fonts are now handled with increased precision internally michael@0: to avoid serious rounding issues if non-integral coordinates are michael@0: encountered. michael@0: michael@0: - Horizontally condensed CFF fonts (using the font matrix) were michael@0: rendered incorrectly. This bug has been introduced after michael@0: release 2.3.5. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - Support for the SFNT cmap 13 table format (as defined by the new michael@0: OpenType 1.6 specification) has been added. michael@0: michael@0: - B/W rasterization of well-hinted TrueType fonts at small sizes michael@0: has been greatly improved. michael@0: michael@0: - Calculation of vertical metrics in OpenType fonts has been michael@0: improved. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - It is now possible to change the emboldening factor in the michael@0: `ftview' demo program with keys `e' and `E'. michael@0: michael@0: - It is now possible to change the slant value in the `ftview' michael@0: demo program with keys `s' and `S'. michael@0: michael@0: - The 5-levels grayscale mode of the `ftraster' module (which michael@0: FreeType doesn't use by default) was broken since version 2.3.0. michael@0: michael@0: - Compilation of the `ftgrays' and `ftraster' modules was broken michael@0: in stand-alone mode. michael@0: michael@0: - Various fixes for compilation on 64bit and 16bit architectures. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.3.9 and 2.3.8 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Very unfortunately, FreeType 2.3.8 contained a change that broke michael@0: its official ABI. The end result is that programs compiled michael@0: against previous versions of the library, but dynamically linked michael@0: to 2.3.8 can experience memory corruption if they call the michael@0: `FT_Get_PS_Font_Info' function. michael@0: michael@0: We recommend all users to upgrade to 2.3.9 as soon as possible, michael@0: or to downgrade to a previous release of the library if this is michael@0: not an option. michael@0: michael@0: The origin of the bug is that a new field was added to the michael@0: publicly defined `PS_FontInfoRec' structure. Unfortunately, michael@0: objects of this type can be stack or heap allocated by callers michael@0: of `FT_Get_PS_Font_Info', resulting in a memory buffer michael@0: overwrite with its implementation in 2.3.8. michael@0: michael@0: If you want to know whether your code is vulnerable to this michael@0: issue, simply search for the substrings `PS_FontInfo' and michael@0: `PS_Font_Info' in your source code. If none is found, your code michael@0: is safe and is not affected. michael@0: michael@0: The FreeType team apologizes for the problem. michael@0: michael@0: - The POSIX support of MacOS resource-fork fonts (Suitcase fonts michael@0: and LaserWriter Type1 PostScript fonts) was broken in 2.3.8. If michael@0: FreeType2 is built without Carbon framework, these fonts are not michael@0: handled correctly. Version 2.3.7 didn't have this bug. michael@0: michael@0: - `FT_Get_Advance' (and `FT_Get_Advances') returned bad values for michael@0: almost all font formats except TrueType fonts. michael@0: michael@0: - Fix a bug in the SFNT kerning table loader/parser which could michael@0: crash the engine if certain malformed tables were encountered. michael@0: michael@0: - Composite SFNT bitmaps are now handled correctly. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - The new functions `FT_Get_CID_Is_Internally_CID_keyed' and michael@0: `FT_Get_CID_From_Glyph_Index' can be used to access CID-keyed michael@0: CFF fonts via CID values. This code has been contributed by michael@0: Michael Toftdal. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - `FT_Outline_Get_InsideBorder' returns FT_STROKER_BORDER_RIGHT michael@0: for empty outlines. This was incorrectly documented. michael@0: michael@0: - The `ftview' demo program now supports UTF-8 encoded strings. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.3.8 and 2.3.7 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - CID-keyed fonts in an SFNT wrapper were not handled correctly. michael@0: michael@0: - The smooth renderer produced truncated images (on the right) for michael@0: outline parts with negative horizontal values. Most fonts don't michael@0: contain outlines left to the y coordinate axis, but the effect michael@0: was very noticeable for outlines processed with FT_Glyph_Stroke, michael@0: using thick strokes. michael@0: michael@0: - `FT_Get_TrueType_Engine_Type' returned a wrong value if both michael@0: configuration macros TT_CONFIG_OPTION_BYTECODE_INTERPRETER and michael@0: TT_CONFIG_OPTION_UNPATENTED_HINTING were defined. michael@0: michael@0: - The `face_index' field in the `FT_Face' structure wasn't michael@0: initialized properly after calling FT_Open_Face and friends with michael@0: a positive face index for CFFs, WinFNTs, and, most importantly, michael@0: for TrueType Collections (TTCs). michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - Rudimentary support for Type 1 fonts and CID-keyed Type 1 fonts michael@0: in an SFNT wrapper has been added -- such fonts are used on the michael@0: Mac. The core SFNT tables `TYP1' and `CID ' are passed to the michael@0: PS Type 1 and CID-keyed PS font drivers; other tables (`ALMX', michael@0: `BBOX', etc.) are not supported yet. michael@0: michael@0: - A new interface to extract advance values of glyphs without michael@0: loading their outlines has been added. The functions are called michael@0: `FT_Get_Advance' and `FT_Get_Advances'; they are defined in file michael@0: `ftadvanc.h' (to be accessed as FT_ADVANCES_H). michael@0: michael@0: - A new function `FT_Get_FSType_Flags' (in FT_FREETYPE_H) has been michael@0: contributed by David Bevan to access the embedding and michael@0: subsetting restriction information of fonts. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - FT_MulFix is now an inlined function; by default, assembler code michael@0: is provided for x86 and ARM. See FT_CONFIG_OPTION_INLINE_MULFIX michael@0: and FT_CONFIG_OPTION_NO_ASSEMBLER (in ftoption.h) for more. michael@0: michael@0: - The handling of `tricky' fonts (this is, fonts which don't work michael@0: with the autohinter, needing the font format's hinting engine) michael@0: has been generalized and changed slightly: michael@0: michael@0: . A new face flag FT_FACE_FLAG_TRICKY indicates that the font michael@0: format's hinting engine is necessary for correct rendering. michael@0: The macro FT_IS_TRICKY can be used to check this flag. michael@0: michael@0: . FT_LOAD_NO_HINTING is now ignored for tricky fonts. To really michael@0: force raw loading of such fonts (without hinting), both michael@0: FT_LOAD_NO_HINTING and FT_LOAD_NO_AUTOHINT must be used -- michael@0: this is something which you probably never want to do. michael@0: michael@0: . Tricky TrueType fonts always use the bytecode interpreter, michael@0: either the patented or unpatented version. michael@0: michael@0: - The function `FT_GlyphSlot_Own_Bitmap' has been moved from michael@0: FT_SYNTHESIS_H to FT_BITMAP_H; it is now part of the `official' michael@0: API. (The functions in FT_SYNTHESIS_H are still subject to michael@0: change, however.) michael@0: michael@0: - In the `ftdiff' demo program you can now toggle the use of michael@0: FT_LOAD_IGNORE_GLOBAL_ADVANCE_WIDTH with key `a'. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.3.7 and 2.3.6 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - If the library was compiled on an i386 platform using gcc, and michael@0: compiler option -O3 was given, `FT_MulFix' sometimes returned michael@0: incorrect results which could have caused problems with michael@0: `FT_Request_Metrics' and `FT_Select_Metrics', returning an michael@0: incorrect descender size. michael@0: michael@0: - Pure CFFs without subfonts were scaled incorrectly if the font michael@0: matrix was non-standard. This bug has been introduced in michael@0: version 2.3.6. michael@0: michael@0: - The `style_name' field in the `FT_FaceRec' structure often michael@0: contained a wrong value for Type 1 fonts. This misbehaviour michael@0: has been introduced in version 2.3.6 while trying to fix michael@0: another problem. [Note, however, that this value is michael@0: informative only since the used algorithm to extract it is michael@0: very simplistic.] michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - Two new macros, FT_OUTLINE_SMART_DROPOUTS and michael@0: FT_OUTLINE_EXCLUDE_STUBS, have been introduced. Together with michael@0: FT_OUTLINE_IGNORE_DROPOUTS (which was ignored previously) it is michael@0: now possible to control the dropout mode of the `raster' module michael@0: (for B&W rasterization), using the `flags' field in the michael@0: `FT_Outline' structure. michael@0: michael@0: - The TrueType bytecode interpreter now passes the dropout mode to michael@0: the B&W rasterizer. This greatly increases the output for small michael@0: ppem values of many fonts like `pala.ttf'. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.3.6 and 2.3.5 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - A bunch of potential security problems have been found. All michael@0: users should update. michael@0: michael@0: - Microsoft Unicode cmaps in TrueType fonts are now always michael@0: preferred over Apple cmaps. This is not a bug per se, but there michael@0: exist some buggy fonts created for MS which have broken Apple michael@0: cmaps. This affects only the automatic selection of FreeType; michael@0: it's always possible to manually select an Apple Unicode cmap if michael@0: desired. michael@0: michael@0: - Many bug fixes to the TrueType bytecode interpreter. michael@0: michael@0: - Improved Mac support. michael@0: michael@0: - Subsetted CID-keyed CFFs are now supported correctly. michael@0: michael@0: - CID-keyed CFFs with subfonts which are scaled in a non-standard michael@0: way are now handled correctly. michael@0: michael@0: - A call to FT_Open_Face with `face_index' < 0 crashed FreeType if michael@0: the font was a Windows (bitmap) FNT/FON. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - The new function `FT_Get_CID_Registry_Ordering_Supplement' gives michael@0: access to those fields in a CID-keyed font. The code has been michael@0: contributed by Derek Clegg. michael@0: michael@0: - George Williams contributed code to validate the new `MATH' michael@0: OpenType table (within the `otvalid' module). The `ftvalid' michael@0: demo program has been extended accordingly. michael@0: michael@0: - An API for cmap 14 support (for Unicode Variant Selectors, UVS) michael@0: has been contributed by George Williams. michael@0: michael@0: - A new face flag FT_FACE_FLAG_CID_KEYED has been added, together michael@0: with a macro FT_IS_CID_KEYED which evaluates to 1 if the font is michael@0: CID-keyed. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - Build support for symbian has been contributed. michael@0: michael@0: - Better WGL4 glyph name support, contributed by Sergey Tolstov. michael@0: michael@0: - Debugging output of the various FT_TRACEX macros is now sent to michael@0: stderr. michael@0: michael@0: - The `ftview' demo program now provides artificial slanting too. michael@0: michael@0: - The `ftvalid' demo program has a new option `-f' to select the michael@0: font index. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.3.5 and 2.3.4 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Some subglyphs in TrueType fonts were handled incorrectly due to michael@0: a missing graphics state reinitialization. michael@0: michael@0: - Large .Z files (as distributed with some X11 packages) weren't michael@0: handled correctly, making FreeType increase the heap stack in an michael@0: endless loop. michael@0: michael@0: - A large number of bugs have been fixed to avoid crashes and michael@0: endless loops with invalid fonts. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - The two new cache functions `FTC_ImageCache_LookupScaler' and michael@0: `FTC_SBit_Cache_LookupScaler' have been added to allow lookup of michael@0: glyphs using an `FTC_Scaler' object; this makes it possible to michael@0: use fractional pixel sizes in the cache. The demo programs have michael@0: been updated accordingly to use this feature. michael@0: michael@0: - A new API `FT_Get_CMap_Format' has been added to get the cmap michael@0: format of a TrueType font. This is useful in handling PDF michael@0: files. The code has been contributed by Derek Clegg. michael@0: michael@0: - The auto-hinter now produces better output by default for michael@0: non-Latin scripts like Indic. This was done by using the CJK michael@0: hinting module as the default instead of the Latin one. Thanks michael@0: to Rahul Bhalerao for this suggestion. michael@0: michael@0: - A new API `FT_Face_CheckTrueTypePatents' has been added to find michael@0: out whether a given TrueType font uses patented bytecode michael@0: instructions. The `ft2demos' bundle contains a new program michael@0: called `ftpatchk' which demonstrates its usage. michael@0: michael@0: - A new API `FT_Face_SetUnpatentedHinting' has been added to michael@0: enable or disable the unpatented hinter. michael@0: michael@0: - Support for Windows FON files in PE format has been contributed michael@0: by Dmitry Timoshkov. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - Vincent Richomme contributed Visual C++ project files for Pocket michael@0: PCs. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.3.4 and 2.3.3 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - A serious bug in the handling of bitmap fonts (and bitmap michael@0: strikes of outline fonts) has been introduced in 2.3.3. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.3.3 and 2.3.2 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Remove a serious regression in the TrueType bytecode interpreter michael@0: that was introduced in version 2.3.2. Note that this does not michael@0: disable the improvements introduced to the interpreter in michael@0: version 2.3.2, only some ill cases that occurred with certain michael@0: fonts (though a few popular ones). michael@0: michael@0: - The auto-hinter now ignores single-point contours for computing michael@0: blue zones. This bug created `wavy' baselines when rendering michael@0: text with various fonts that use these contours to model michael@0: mark-attach points (these are points that are never rasterized michael@0: and are placed outside of the glyph's real outline). michael@0: michael@0: - The `rsb_delta' and `lsb_delta' glyph slot fields are now set to michael@0: zero for mono-spaced fonts. Otherwise code that uses them would michael@0: essentially ruin the fixed-advance property. michael@0: michael@0: - Fix CVE-2007-1351 which can cause an integer overflow while michael@0: parsing BDF fonts, leading to a potentially exploitable heap michael@0: overflow condition. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - Fixed compilation issues on some 64-bit platforms (see ChangeLog michael@0: for details). michael@0: michael@0: - A new demo program `ftdiff' has been added to compare TrueType michael@0: hinting, FreeType's auto hinting, and rendering without hinting michael@0: in three columns. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.3.2 and 2.3.1 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - FreeType returned incorrect kerning information from TrueType michael@0: fonts when the bytecode interpreter was enabled. This happened michael@0: due to a typo introduced in version 2.3.0. michael@0: michael@0: - Negative kerning values from PFM files are now reported michael@0: correctly (they were read as 16-bit unsigned values from the michael@0: file). michael@0: michael@0: - Fixed a small memory leak when `FT_Init_FreeType' failed for michael@0: some reason. michael@0: michael@0: - The Postscript hinter placed and sized very thin and ghost stems michael@0: incorrectly. michael@0: michael@0: - The TrueType bytecode interpreter has been fixed to get rid of michael@0: most of the rare differences seen in comparison to the Windows michael@0: font loader. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - The auto-hinter now better deals with serifs and corner cases michael@0: (e.g., glyph '9' in Arial at 9pt, 96dpi). It also improves michael@0: spacing adjustments and doesn't change widths for non-spacing michael@0: glyphs. michael@0: michael@0: - Many Mac-specific functions are deprecated (but still michael@0: available); modern replacements have been provided for them. michael@0: See the documentation in file `ftmac.h'. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.3.1 and 2.3.0 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - The TrueType interpreter sometimes returned incorrect horizontal michael@0: metrics due to a bug in the handling of the SHZ instruction. michael@0: michael@0: - A typo in a security check introduced after version 2.2.1 michael@0: prevented FreeType to render some glyphs in CFF fonts. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.3.0 and 2.2.1 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - The PCF font loader is now much more robust while loading michael@0: malformed font files. michael@0: michael@0: - Various memory leaks have been found and fixed. michael@0: michael@0: - The TrueType name loader now deals properly with some fonts that michael@0: encode their names in UTF-16 (the specification was vague, and michael@0: the code incorrectly assumed UCS-4). michael@0: michael@0: - Fixed the TrueType bytecode loader to deal properly with subtle michael@0: monochrome/gray issues when scaling the CVT. Some fonts michael@0: exhibited bad rendering artifacts otherwise. michael@0: michael@0: - `FT_GlyphSlot_Embolden' now supports vertical layouts correctly michael@0: (it mangled the vertical advance height). michael@0: michael@0: - Fixed byte endian issues of `ftmac.c' to support Mac OS X on michael@0: i386. michael@0: michael@0: - The PFR font loader no longer erroneously tags font files michael@0: without any outlines as FT_FACE_FLAG_SCALABLE. michael@0: michael@0: michael@0: II. NEW API FUNCTIONS michael@0: michael@0: - `FT_Library_SetLcdFilter' allows you to select a special filter michael@0: to be applied to the bitmaps generated by `FT_Render_Glyph' if michael@0: one of the FT_RENDER_MODE_LCD and FT_RENDER_MODE_LCD_V modes has michael@0: been selected. This filter is used to reduce color fringes; michael@0: several settings are available through the FT_LCD_FILTER_XXX michael@0: enumeration. michael@0: michael@0: Its declaration and documentation can be found in file michael@0: `include/freetype/ftlcdfil.h' (to be accessed with macro michael@0: FT_LCD_FILTER_H). michael@0: michael@0: *IMPORTANT*: This function returns an error michael@0: (FT_Err_Unimplemented_Feature) in default builds of the library michael@0: for patent reasons. See below. michael@0: michael@0: - `FT_Get_Gasp' allows you to query the flags of the TrueType michael@0: `gasp' table for a given character pixel size. This is useful michael@0: to duplicate the text rendering of MS Windows when the native michael@0: bytecode interpreter is enabled (which isn't the default for michael@0: other patent reasons). michael@0: michael@0: Its declaration and documentation can be found in file michael@0: `include/freetype/ftgasp.h' (to be accessed with macro michael@0: FT_GASP_H). michael@0: michael@0: michael@0: III. IMPORTANT CHANGES michael@0: michael@0: - The auto-hinter has been tuned a lot to improve its results with michael@0: serif fonts, resulting in much better font rendering of many web michael@0: pages. michael@0: michael@0: - The unpatented hinter is now part of the default build of the michael@0: library; we have added code to automatically support `tricky' michael@0: fonts that need it. michael@0: michael@0: This means that FreeType should `just work' with certain Asian michael@0: fonts, like MingLiU, which cannot properly be loaded without a michael@0: bytecode interpreter, but which fortunately do not use any of michael@0: the patented bytecode opcodes. We detect these fonts by name, michael@0: so please report any font file that doesn't seem to work with michael@0: FreeType, and we shall do what we can to support it in a next michael@0: release. michael@0: michael@0: Note that the API hasn't changed, so you can still force michael@0: unpatented hinting with a special parameter to `FT_Open_Face' as michael@0: well. This might be useful in same cases; for example, a PDF michael@0: reader might present a user option to activate it to deal with michael@0: certain `tricky' embedded fonts which cannot be clearly michael@0: identified. michael@0: michael@0: If you are a developer for embedded systems, you might want to michael@0: *disable* the feature to save code space by undefining michael@0: TT_CONFIG_OPTION_UNPATENTED_HINTING in file `ftoption.h'. michael@0: michael@0: - LCD-optimized rendering is now *disabled* in all default builds michael@0: of the library, mainly due to patent issues. For more michael@0: information see: michael@0: michael@0: http://lists.gnu.org/archive/html/freetype/2006-09/msg00064.html michael@0: michael@0: A new configuration macro FT_CONFIG_OPTION_SUBPIXEL_RENDERING michael@0: has been introduced in `ftoption.h'; manually define it in this michael@0: file if you want to re-enable the feature. michael@0: michael@0: The change only affects the implementation, not the FreeType michael@0: API. This means that clients don't need to be modified, because michael@0: the library still generates LCD decimated bitmaps, but with the michael@0: added constraint that R=G=B on each triplet. michael@0: michael@0: The displayed result should be equal to normal anti-aliased michael@0: rendering. michael@0: michael@0: Additionally, if FT_CONFIG_OPTION_SUBPIXEL_RENDERING is not michael@0: defined, the new `FT_Library_SetLcdFilter' function returns the michael@0: FT_Err_Unimplemented_Feature error code. michael@0: michael@0: - Some computation bugs in the TrueType bytecode interpreter were michael@0: found, which allow us to get rid of very subtle and rare michael@0: differences we had experienced with the Windows renderer. michael@0: michael@0: - It is now possible to cross-compile the library easily. See the michael@0: file `docs/INSTALL.CROSS' for details. michael@0: michael@0: - The file `src/base/ftmac.c' now contains code for Mac OS X only; michael@0: its deprecated function `FT_GetFile_From_Mac_Font_Name' always michael@0: returns an error even if the QuickDraw framework is available. michael@0: The previous version has been moved to `builds/mac/ftmac.c'. michael@0: michael@0: Selecting configure option `--with-quickdraw-carbon' makes the michael@0: build process use the original `ftmac.c' file instead of the Mac michael@0: OS X-only version. michael@0: michael@0: michael@0: IV. MISCELLANEOUS michael@0: michael@0: - Various performance and memory footprint optimizations have been michael@0: performed on the TrueType and CFF font loaders, sometimes with michael@0: very drastic benefits (e.g., the TrueType loader is now about michael@0: 25% faster; FreeType should use less heap memory under nearly michael@0: all conditions). michael@0: michael@0: - The anti-aliased rasterizer has been optimized and is now 15% to michael@0: 25% percent faster than in previous versions, depending on michael@0: content. michael@0: michael@0: - The Type 1 loader has been improved; as an example, it now skips michael@0: top-level dictionaries properly. michael@0: michael@0: - Better support for Mac fonts on POSIX systems, plus compilation michael@0: fixes for Mac OS X on ppc64 where `ftmac.c' cannot be built. michael@0: michael@0: - Configuration without `--with-old-mac-fonts' does not include michael@0: `ftmac.c' (this was the behaviour in FreeType version 2.1.10). michael@0: michael@0: - The TrueTypeGX validator (gxvalid) checks the order of glyph IDs michael@0: in the kern table. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.2.1 and 2.2 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Various integer overflows have been fixed. michael@0: michael@0: - PFB fonts with MacOS resource fork weren't handled correctly on michael@0: non-MacOS platforms. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.2 and 2.1.10 michael@0: michael@0: (not released officially) michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Vertical metrics for SFNT fonts were incorrect sometimes. michael@0: michael@0: - The FT_HAS_KERNING macro always returned 0. michael@0: michael@0: - CFF OpenType fonts didn't return correct vertical metrics for michael@0: glyphs with outlines. michael@0: michael@0: - If FreeType was compiled without hinters, all font formats based michael@0: on PS outlines weren't scaled correctly. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - Version 2.2 no longer exposes its internals, this is, the header michael@0: files located in the `include/freetype/internal' directory of michael@0: the source package are not copied anymore by the `make install' michael@0: command. Consequently, a number of rogue clients which directly michael@0: access FreeType's internal functions and structures won't michael@0: compile without modification. michael@0: michael@0: We provide patches for most of those rogue clients. See the michael@0: following page for more information: michael@0: michael@0: http://www.freetype.org/freetype2/patches/rogue-patches.html michael@0: michael@0: Note that, as a convenience to our Unix desktop users, version michael@0: 2.2 is *binary* compatible with FreeType 2.1.7, which means that michael@0: installing this release on an existing distribution shall not michael@0: break any working desktop. michael@0: michael@0: - FreeType's build mechanism has been redesigned. With GNU make michael@0: it is now sufficient in most cases to edit two files: michael@0: `modules.cfg', to select the library components, and the michael@0: configuration file `include/freetype/config/ftoption.h' (which michael@0: can be copied to the objects directory). Removing unused module michael@0: directories to prevent its compilation and editing michael@0: `include/freetype/config/ftmodule.h' is no longer necessary. michael@0: michael@0: - The LIGHT hinting algorithm produces more pleasant results. michael@0: Also, using the FT_LOAD_TARGET_LIGHT flags within FT_Load_Glyph michael@0: always forces auto-hinting, as a special exception. This allows michael@0: you to experiment with it even if you have enabled the TrueType michael@0: bytecode interpreter in your build. michael@0: michael@0: - The auto hinter now employs a new algorithm for CJK fonts, based michael@0: on Akito Hirai's patch. Note that this only works for fonts michael@0: with a Unicode charmap at the moment. michael@0: michael@0: - The following callback function types have changed slightly (by michael@0: adding the `const' keyword where appropriate): michael@0: michael@0: FT_Outline_MoveToFunc michael@0: FT_Outline_LineToFunc michael@0: FT_Outline_ConicToFunc michael@0: FT_Outline_CubicToFunc michael@0: FT_SpanFunc michael@0: FT_Raster_RenderFunc michael@0: michael@0: FT_Glyph_TransformFunc michael@0: FT_Renderer_RenderFunc michael@0: FT_Renderer_TransformFunc michael@0: michael@0: Note that this doesn't affect binary backward compatibility. michael@0: michael@0: - On MacOS, new APIs have been added as replacements for legacy michael@0: APIs: `FT_New_Face_From_FSRef' for `FT_New_Face_From_FSSpec', michael@0: and `FT_GetFile_From_Mac_ATS_Name' for michael@0: `FT_GetFile_From_Mac_Name'. Legacy APIs are still available, if michael@0: FreeType is built without disabling them. michael@0: michael@0: - A new API `FT_Select_Size' has been added to select a bitmap michael@0: strike by its index. Code using other functions to select michael@0: bitmap strikes should be updated to use this function. michael@0: michael@0: - A new API `FT_Get_SubGlyph_Info' has been added to retrieve michael@0: subglyph data. This can be used by rogue clients which used to michael@0: access the internal headers to get the corresponding data. michael@0: michael@0: - In 2.1.10, the behaviour of `FT_Set_Pixel_Sizes' was changed for michael@0: BDF/PCF fonts, and only for them. This causes inconsistency. michael@0: In this release, we undo the change. The intent of the change michael@0: in 2.1.10 is to allow size selection through real dimensions, michael@0: which can now be done through `FT_Request_Size'. michael@0: michael@0: - Some security issues were discovered and fixed in the CFF and michael@0: Type 1 loader, causing crashes of FreeType by malformed font michael@0: files. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - The documentation for FT_LOAD_TARGET_XXX and FT_RENDER_MODE_XXX michael@0: values now better reflects its usage and differences: One set is michael@0: used to specify the hinting algorithm, the other to specify the michael@0: pixel rendering mode. michael@0: michael@0: - `FT_New_Face' and `FT_New_Face_From_FSSpec' in ftmac.c have been michael@0: changed to count supported scalable faces (sfnt, LWFN) only, and michael@0: to return the number of available faces via face->num_faces. michael@0: Unsupported bitmap faces (fbit, NFNT) are ignored. michael@0: michael@0: - builds/unix/configure has been improved for MacOS X. It now michael@0: automatically checks available functions in Carbon library, and michael@0: prepare to use newest functions by default. Options to specify michael@0: the dependencies of each Carbon APIs (FSSpec, FSRef, old/new michael@0: QuickDraw, ATS) are available too. By manual disabling of all michael@0: QuickDraw functionality, FreeType can be built without michael@0: `deprecated function' warnings on MacOS 10.4.x, but michael@0: FT_GetFile_Mac_Name in ftmac.c then is changed to a dummy michael@0: function, and returns an `unimplemented' error. For details see michael@0: builds/mac/README. michael@0: michael@0: - SFNT cmap handling has been improved, mainly to run much faster michael@0: with CJK fonts. michael@0: michael@0: - A new function `FT_Get_TrueType_Engine_Type (declared in michael@0: `FT_MODULE_H') is provided to determine the status of the michael@0: TrueType bytecode interpreter compiled into the library michael@0: (patented, unpatented, unimplemented). michael@0: michael@0: - Vertical metrics of glyphs are synthesized if the font does not michael@0: provide such information. You can tell whether the metrics are michael@0: synthesized or not by checking the FT_FACE_FLAG_VERTICAL flag of michael@0: the face. michael@0: michael@0: - The demo programs `ftview' and `ftstring' have been rewritten michael@0: for better readability. `ftview' has a new switch `-p' to test michael@0: FT_New_Memory_Face (instead of FT_New_Face). michael@0: michael@0: - FreeType now honours bit 1 in the `head' table of TrueType fonts michael@0: (meaning `left sidebearing point at x=0'). This helps with some michael@0: buggy fonts. michael@0: michael@0: - Rudimentary support for Adobe's new `SING Glyphlet' format. See michael@0: michael@0: http://www.adobe.com/products/indesign/sing_gaiji.html michael@0: michael@0: for more information. michael@0: michael@0: - The `ftdump' program from the `ft2demos' bundle now shows some michael@0: information about charmaps. It also supports a new switch `-v' michael@0: to increase verbosity. michael@0: michael@0: - Better AFM support. This includes track kerning support. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.1.10 and 2.1.9 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - The size comparison for BDF and PCF files could fail sometimes. michael@0: michael@0: - Some CFF files were still not loaded correctly. Patch from michael@0: Derek Noonburg. michael@0: michael@0: - The stroker still had some serious bugs. michael@0: michael@0: - Boris Letocha fixed a bug in the TrueType interpreter: The michael@0: NPUSHW instruction wasn't skipped correctly in IF clauses. Some michael@0: fonts like `Helvetica 75 Bold' failed. michael@0: michael@0: - Another serious bug in handling TrueType hints caused many michael@0: distortions. It has been introduced in version 2.1.8, and it is michael@0: highly recommended to upgrade. michael@0: michael@0: - FreeType didn't properly parse empty Type 1 glyphs. michael@0: michael@0: - An unbound dynamic buffer growth was fixed in the PFR loader. michael@0: michael@0: - Several bugs have been fixed in the cache sub-system. michael@0: michael@0: - FreeType behaved incorrectly when resizing two distinct but very michael@0: close character pixel sizes through `FT_Set_Char_Size' (Savannah michael@0: bug #12263). michael@0: michael@0: - The auto-hinter didn't work properly for fonts without a Unicode michael@0: charmap -- it even refused to load the glyphs. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - Many fixes have been applied to drastically reduce the amount of michael@0: heap memory used by FreeType, especially when using michael@0: memory-mapped font files (which is the default on Unix systems michael@0: which support them). michael@0: michael@0: - The auto-hinter has been replaced with a new module, called the michael@0: `auto-fitter'. It consumes less memory than its predecessor, michael@0: and it is prepared to support non-latin scripts better in next michael@0: releases. michael@0: michael@0: - George Williams contributed code to read kerning data from PFM michael@0: files. michael@0: michael@0: - FreeType now uses the TT_NAME_ID_PREFERRED_FAMILY and michael@0: TT_NAME_ID_PREFERRED_SUBFAMILY strings (if available) for michael@0: setting family and style in SFNT fonts (patch from Kornfeld michael@0: Eliyahu Peter). michael@0: michael@0: - A new API `FT_Sfnt_Table_Info' (in FT_TRUETYPE_TABLES_H) has michael@0: been added to retrieve name and size information of SFNT tables. michael@0: michael@0: - A new API `FT_OpenType_Validate' (in FT_OPENTYPE_VALIDATE_H) has michael@0: been added to validate OpenType tables (BASE, GDEF, GPOS, GSUB, michael@0: JSTF). After validation it is no longer necessary to check michael@0: for errors in those tables while accessing them. michael@0: michael@0: Note that this module might be moved to another library in the michael@0: future to avoid a tight dependency between FreeType and the michael@0: OpenType specification. michael@0: michael@0: - A new API in FT_BITMAP_H (`FT_Bitmap_New', `FT_Bitmap_Convert', michael@0: `FT_Bitmap_Copy', `FT_Bitmap_Embolden', `FT_Bitmap_Done') has michael@0: been added. Its use is to convert an FT_Bitmap structure in michael@0: 1bpp, 2bpp, 4bpp, or 8bpp format into another 8bpp FT_Bitmap, michael@0: probably using a different pitch, and to further manipulate it. michael@0: michael@0: - A new API `FT_Outline_Embolden' (in FT_OUTLINE_H) gives finer michael@0: control how outlines are embolded. michael@0: michael@0: - `FT_GlyphSlot_Embolden' (in FT_SYNTHESIS_H) now handles bitmaps michael@0: also (code contributed by Chia I Wu). Note that this function michael@0: is still experimental and may be replaced with a better API. michael@0: michael@0: - The method how BDF and PCF bitmap fonts are accessed has been michael@0: refined. Formerly, FT_Set_Pixel_Sizes and FT_Set_Char_Size michael@0: were synonyms in FreeType's BDF and PCF interface. This has michael@0: changed now. FT_Set_Pixel_Sizes should be used to select the michael@0: actual font dimensions (the `strike', which is the sum of the michael@0: `FONT_ASCENT' and `FONT_DESCENT' properties), while michael@0: FT_Set_Char_Size selects the `nominal' size (the `PIXELSIZE' michael@0: property). In both functions, the width parameter is ignored. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - The BDF driver no longer converts all returned bitmaps with a michael@0: depth of 2bpp or 4bpp to a depth of 8bpp. The documentation has michael@0: not mentioned this explicitly, but implementors might have michael@0: relied on this after looking into the source files. michael@0: michael@0: - A new option `--ftversion' has been added to freetype-config to michael@0: return the FreeType version. michael@0: michael@0: - The memory debugger has been updated to dump allocation michael@0: statistics on all allocation sources in the library. This is michael@0: useful to spot greedy allocations when loading and processing michael@0: fonts. michael@0: michael@0: - We removed a huge array of constant pointers to constant strings michael@0: in the `psnames' module. The problem was that compilations in michael@0: PIC mode (i.e., when generating a Unix shared object/dll) put michael@0: the array into the non-shared writable section of the library michael@0: since absolute pointers are not relocatable by nature. michael@0: michael@0: This reduces the memory consumption by approximately 16KByte per michael@0: process linked to FreeType. We now also store the array in a michael@0: compressed form (as a trie) which saves about 20KByte of code as michael@0: well. michael@0: michael@0: - Kirill Smelkov provided patches to make src/raster/ftraster.c michael@0: compile stand-alone again. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.1.9 and 2.1.8 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - The function `FT_Get_CharMap_Index' was only declared, without michael@0: any real code. For consistency, it has been renamed to michael@0: `FT_Get_Charmap_Index'. (This function is needed to implement michael@0: cmap caches.) michael@0: michael@0: - `FT_Outline_Get_BBox' sometimes returned incorrect values for michael@0: conic outlines (e.g., for TrueType fonts). michael@0: michael@0: - Handling of `bhed' table has been fixed. michael@0: michael@0: - The TrueType driver with enabled byte code interpreter sometimes michael@0: returned artifacts due to incorrect rounding. This bug has been michael@0: introduced after version 2.1.4. michael@0: michael@0: - The BDF driver dropped the last glyph in the font. michael@0: michael@0: - The BDF driver now uses the DEFAULT_CHAR property (if available) michael@0: to select a glyph shape for the undefined glyph. michael@0: michael@0: - The stroker failed for closed outlines and single points. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - George Williams contributed code to handle Apple's font michael@0: distortion technology found in GX fonts (`avar', `cvar', `fvar', michael@0: and `gvar' tables; the Multiple Masters API has been slightly michael@0: extended to cope with the new functionality). michael@0: michael@0: - The `FT_GlyphSlotRec' structure has been extended: The elements michael@0: `lsb_delta' and `rsb_delta' give the difference between hinted michael@0: and unhinted left and right side bearings if autohinting is michael@0: active. Using those values can improve the inter-letter spacing michael@0: considerably. See the documentation of `FT_GlyphSlotRec' and michael@0: the `ftstring' demo program how to use it. michael@0: michael@0: - Loading TrueType and Type 1 fonts has been made much faster. michael@0: michael@0: - The stroker is no longer experimental (but the cache subsystem michael@0: still is). michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - A new documentation file `formats.txt' describes various font michael@0: formats supported (and not supported) by FreeType. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.1.8 and 2.1.7 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - The native TrueType hinter contained some bugs which prevented michael@0: some fonts to be rendered correctly, most notably Legendum.otf. michael@0: michael@0: - The PostScript hinter now produces improved results. michael@0: michael@0: - The linear advance width and height values were incorrectly michael@0: rounded, making them virtually unusable if not loaded with michael@0: FT_LOAD_LINEAR_DESIGN. michael@0: michael@0: - Indexing CID-keyed CFF fonts is now working: The glyph index is michael@0: correctly treated as a CID, similar to FreeType's CID driver michael@0: module. Note that CID CMap support is still missing. michael@0: michael@0: - The FT_FACE_FLAGS_GLYPH_NAMES flag is now set correctly for all michael@0: font formats. michael@0: michael@0: - Some subsetted Type 1 fonts weren't parsed correctly. This bug michael@0: has been introduced in 2.1.7. In summary, the Type 1 parser has michael@0: become more robust. michael@0: michael@0: - Non-decimal numbers weren't parsed correctly in PS fonts. michael@0: michael@0: - The WinFNT driver now correctly reports FT_ENCODING_NONE for all michael@0: but one encoding. Use the new FT_WinFNT_ID_XXX values together michael@0: with `FT_Get_WinFNT_Header' to get the WinFNT charset ID. michael@0: michael@0: - The descender metrics (face->size->metrics.descender) for WinFNT michael@0: bitmap fonts had the wrong sign. michael@0: michael@0: - The (emulated) `seac' support for CFF fonts was broken. michael@0: michael@0: - The `flex' operator didn't work for CFF fonts. michael@0: michael@0: - PS glyphs which use the `hintmask' operator haven't been michael@0: rendered correctly in some cases. michael@0: michael@0: - Metrics for BDF and PCF bitmap font formats have been fixed. michael@0: michael@0: - Autohinting is now disabled for glyphs which are vertically michael@0: distorted or mirrored (using a transformation matrix). This michael@0: fixes a bug which produced zero-height glyphs. michael@0: michael@0: - The `freetype-config' script now handles --prefix and michael@0: --exec-prefix correctly; it also returns the proper --rpath (or michael@0: -R) value if FreeType has been built as a shared library. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - Both PCF and BDF drivers now handle the SETWIDTH_NAME and michael@0: ADD_STYLE_NAME properties. Values are appended to michael@0: face->style_name; example: `Bold SemiCondensed'. michael@0: michael@0: - The PCF driver now handles bitmap fonts compressed with the LZW michael@0: algorithm (extension .pcf.Z, compressed with `compress'). michael@0: michael@0: - A new API function `FT_Get_CMap_Language_ID' (declared in michael@0: `tttables.h') is available to get the language ID of a michael@0: TrueType/SFNT cmap. michael@0: michael@0: - The hexadecimal format of data after the `StartData' command in michael@0: CID-keyed Type 1 fonts is now supported. While this can't occur michael@0: in file-based fonts, it can happen in document-embedded michael@0: resources of PostScript documents. michael@0: michael@0: - Embedded bitmaps in SFNT-based CFF fonts are now supported. michael@0: michael@0: - A simple API is now available to control FreeType's tracing michael@0: mechanism if compiled with FT_DEBUG_LEVEL_TRACE. See the file michael@0: `ftdebug.h' for more details. michael@0: michael@0: - YAMATO Masatake contributed improved handling of MacOS resource michael@0: forks on non-MacOS platforms (for example, Linux can mount MacOS michael@0: file systems). michael@0: michael@0: - Support for MacOS has been improved; there is now a new function michael@0: `FT_New_Face_From_FSSpec' similar to `FT_New_Face' except that michael@0: it accepts an FSSpec instead of a path. michael@0: michael@0: - The cache sub-system has been rewritten. michael@0: michael@0: - There is now support for deinstallation of faces. michael@0: michael@0: - A new API function `FTC_Manager_RemoveFaceID' has been added michael@0: to delete all `idle' nodes that correspond to a given michael@0: FTC_FaceID. All `locked' nodes (i.e., those with a reference michael@0: count > 0), will be modified to prevent them from appearing in michael@0: further lookups (they will be cleaned normally when their michael@0: reference count reaches 0). michael@0: michael@0: - There is now support for point scaling (i.e., providing michael@0: character sizes in points + dpis, instead of pixels). michael@0: michael@0: - Three abstract cache classes are now available: michael@0: michael@0: FTC_GCache: Used to store one glyph item per cache node, michael@0: with the ability to group common attributes into michael@0: `families'. This replaces the old michael@0: FTC_GlyphCache class. michael@0: michael@0: FTC_ICache: Used to store one FT_Glyph per cache node. This michael@0: extends FTC_GCache. Family definition, family michael@0: comparison, and glyph loading are however left michael@0: to sub-classes. michael@0: michael@0: FTC_SCache: Used to store up to 16 small bitmaps per cache michael@0: node. This extends FTC_GCache. Family michael@0: definition, family comparison and glyph loading michael@0: are however left to sub-classes. michael@0: michael@0: - The file `src/cache/ftcbasic.c' implements: michael@0: michael@0: FTC_ImageCache: Extends FTC_ICache; implements family michael@0: definitions and glyph loading similar to the michael@0: old API. michael@0: michael@0: FTC_SBitCache: Extends FTC_SCache, implements family michael@0: definitions and glyph loading similar to the michael@0: old API michael@0: michael@0: Client applications should be able to extend FTC_GCache, michael@0: FTC_ICache, or FTC_SCache much more easily (i.e., less code to michael@0: write, and less callbacks). For example, one could envision michael@0: caches that are capable of storing transformed (obliqued), michael@0: stroked, emboldened, or colored glyph images. Use michael@0: `ftcbasic.c' as an example. michael@0: michael@0: - All public APIs are now in `include/freetype/ftcache.h', (to michael@0: be accessed as `FT_CACHE_H'). The contents of michael@0: `include/freetype/cache/' is only needed by applications that michael@0: wish to implement their own caches. michael@0: michael@0: - There were some major performance improvements through the use michael@0: of various programming tricks. Cache hits are up to 70% michael@0: faster than in the old code. michael@0: michael@0: - The FTC_CMapCache has been simplified. Charmaps can only be michael@0: accessed by index right now. There is also a new API named michael@0: `FT_Charmap_GetIndex' for this purpose. michael@0: michael@0: - The demo programs have been updated to the new code. The michael@0: previous versions will not work with the current one. michael@0: michael@0: - Using an invalid face index in FT_Open_Face and friends now michael@0: causes an error even if the font contains a single face only. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - Wolfgang Domröse contributed support files for building FreeType michael@0: on the Atari using the PureC compiler. Note that the Atari is a michael@0: 16bit platform. michael@0: michael@0: - Vitaliy Pasternak contributed project files for VS.NET 2003. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.1.7 and 2.1.6 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Updated to newest libtool version, fixing build problems on michael@0: various platforms. michael@0: michael@0: - On Unix platforms, `make install' didn't copy the correct michael@0: `ftconfig.h' file. michael@0: michael@0: Note that version 2.1.7 contains the same library C source code as michael@0: version 2.1.6. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.1.6 and 2.1.5 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - The PFR font driver didn't load kerning tables correctly, and michael@0: the functions in FT_PFR_H didn't work at all. michael@0: michael@0: - Type 1 font files in binary format (PFB) with an end-of-file michael@0: indicator weren't accepted by the FreeType engine. michael@0: michael@0: - Fonts which contain /PaintType and /StrokeWidth no longer cause michael@0: a segfault. This bug has been introduced in version 2.1.5. michael@0: michael@0: - Fonts loaded with FT_LOAD_RENDER no longer cause strange michael@0: results. This bug has been introduced in version 2.1.5. michael@0: michael@0: - Some Windows (bitmap) FNT/FON files couldn't be handled michael@0: correctly. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - The internal module API has been heavily changed in favor of michael@0: massive simplifications within the font engine. This also means michael@0: that authors of third-party modules must adapt their code to the michael@0: new scheme. michael@0: michael@0: NOTE: THE NEW SCHEME IS NOT COMPLETED YET. PLEASE WAIT UNTIL A michael@0: FINAL ANNOUNCEMENT! michael@0: michael@0: - The PostScript parser has been enhanced to handle comments and michael@0: strings correctly. Additionally, more syntax forms are michael@0: recognized. michael@0: michael@0: - Added the optional unpatented hinting system for TrueType. It michael@0: allows typefaces which need hinting to produce correct glyph michael@0: forms (e.g., Chinese typefaces from Dynalab) to work acceptably michael@0: without infringing Apple patents. This system is compiled only michael@0: if TT_CONFIG_OPTION_COMPILE_UNPATENTED_HINTING is defined in michael@0: ftoption.h (activated by default). michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - There is now a guard in the public header files to protect michael@0: against inclusion of freetype.h from FreeType 1. michael@0: michael@0: - Direct inclusion of freetype.h and other public header files no michael@0: longer works. You have to use the documented scheme michael@0: michael@0: #include michael@0: #include FT_FREETYPE_H michael@0: michael@0: to load freetype.h with a symbolic name. This protects against michael@0: renaming of public header files (which shouldn't happen but michael@0: actually has, avoiding two public header files with the same michael@0: name). michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.1.5 and 2.1.4 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Parsing the /CIDFontName field now removes the leading slash to michael@0: be in sync with other font drivers. michael@0: michael@0: - gzip support was buggy. Some fonts could not be read. michael@0: michael@0: - Fonts which have nested subglyphs more than one level deep no michael@0: longer cause a segfault. michael@0: michael@0: - Creation of synthetic cmaps for fonts in CFF format was broken michael@0: partially. michael@0: michael@0: - Numeric font dictionary entries for synthetic fonts are no michael@0: longer overwritten. michael@0: michael@0: - The font matrix wasn't applied to the advance width for Type1, michael@0: CID, and CFF fonts. This caused problems when loading certain michael@0: synthetic Type 1 fonts like `Helvetica Narrow'. michael@0: michael@0: - The test for the charset registry in BDF and PCF fonts is now michael@0: case-insensitive. michael@0: michael@0: - FT_Vector_Rotate sometimes returned strange values due to michael@0: rounding errors. michael@0: michael@0: - The PCF driver now returns the correct number of glyphs michael@0: (including an artificial `notdef' glyph at index 0). michael@0: michael@0: - FreeType now supports buggy CMaps which are contained in many michael@0: CJK fonts from Dynalab. michael@0: michael@0: - Opening an invalid font on a Mac caused a segfault due to michael@0: double-freeing memory. michael@0: michael@0: - BDF fonts with more than 32768 glyphs weren't supported michael@0: properly. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - Accessing bitmap font formats has been synchronized. To do that michael@0: the FT_Bitmap_Size structure has been extended to contain new michael@0: fields `size', `x_ppem', and `y_ppem'. michael@0: michael@0: - The FNT driver now returns multiple faces, not multiple strikes. michael@0: michael@0: - The `psnames' module has been updated to the Adobe Glyph List michael@0: version 2.0. michael@0: michael@0: - The `psnames' module now understands `uXXXX[X[X]]' glyph names. michael@0: michael@0: - The algorithm for guessing the font style has been improved. michael@0: michael@0: - For fonts in SFNT format, root->height is no longer increased if michael@0: the line gap is zero. There exist fonts (containing e.g. form michael@0: drawing characters) which intentionally have a zero line gap michael@0: value. michael@0: michael@0: - ft_glyph_bbox_xxx flags are now deprecated in favour of michael@0: FT_GLYPH_BBOX_XXX. michael@0: michael@0: - ft_module_xxx flags are now deprecated in favour of michael@0: FT_MODULE_XXX. michael@0: michael@0: - FT_ENCODING_MS_{SJIS,GB2312,BIG5,WANSUNG,JOHAB} are now michael@0: deprecated in favour of michael@0: FT_ENCODING_{SJIS,GB2312,GIB5,WANSONG,JOHAB} -- those encodings michael@0: are not specific to Microsoft. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - The autohinter has been further improved; for example, `m' michael@0: glyphs now retain its vertical symmetry. michael@0: michael@0: - Partial support of Mac fonts on non-Mac platforms. michael@0: michael@0: - `make refdoc' (after first `make') builds the HTML michael@0: documentation. You need Python for this. michael@0: michael@0: - The make build system should now work more reliably on DOS-like michael@0: platforms. michael@0: michael@0: - Support for EMX gcc and Watson C/C++ compilers on MS-DOS has michael@0: been added. michael@0: michael@0: - Better VMS build support. michael@0: michael@0: - Support for the pkg-config package by providing a `freetype.pc' michael@0: file. michael@0: michael@0: - New configure option --with-old-mac-fonts for Darwin. michael@0: michael@0: - Some source files have been renamed (mainly to fit into the 8.3 michael@0: naming scheme). michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.1.4 and 2.1.3 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Updated to newest libtool version, fixing build problems on michael@0: various platforms. michael@0: michael@0: - A fix in the Gzip stream reader: It couldn't read certain .gz michael@0: files properly due to a small typo. In certain cases, FreeType michael@0: could also loop endlessly when trying to load tiny gzipped michael@0: files. michael@0: michael@0: - The configure script now tries to use the system-wide zlib when michael@0: it finds one (instead of the copy found in src/gzip). And michael@0: `freetype-config' has been updated to return relevant flags in michael@0: this case when invoked with `--libs' (e.g. `-lzlib'). michael@0: michael@0: - Certain fonts couldn't be loaded by 2.1.3 because they lacked a michael@0: Unicode charmap (e.g. SYMBOL.TTF). FreeType erroneously michael@0: rejected them. michael@0: michael@0: - The CFF loader was modified to accept fonts which only contain a michael@0: subset of their reference charset. This prevented the correct michael@0: use of PDF-embedded fonts. michael@0: michael@0: - The logic to detect Unicode charmaps has been modified. This is michael@0: required to support fonts which include both 16-bit and 32-bit michael@0: charmaps (like very recent asian ones) using the new 10 and 12 michael@0: SFNT formats. michael@0: michael@0: - The TrueType loader now limits the depth of composite glyphs. michael@0: This is necessary to prevent broken fonts to break the engine by michael@0: blowing the stack with recursive glyph definitions. michael@0: michael@0: - The CMap cache is now capable of managing UCS-4 character codes michael@0: that are mapped through extended charmaps in recent michael@0: TrueType/OpenType fonts. michael@0: michael@0: - The cache sub-system now properly manages out-of-memory michael@0: conditions instead of blindly reporting them to the caller. michael@0: This means that it will try to empty the cache before restarting michael@0: its allocations to see if that can help. michael@0: michael@0: - The PFR driver didn't return the list of available embedded michael@0: bitmaps properly. michael@0: michael@0: - There was a nasty memory leak when using embedded bitmaps in michael@0: certain font formats. michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - David Chester contributed some enhancements to the auto-hinter michael@0: that significantly increase the quality of its output. The michael@0: Postscript hinter was also improved in several ways. michael@0: michael@0: - The FT_RENDER_MODE_LIGHT render mode was implemented. michael@0: michael@0: - A new API function called `FT_Get_BDF_Property' has been added michael@0: to FT_BDF_H to retrieve BDF properties from BDF _and_ PCF font michael@0: files. THIS IS STILL EXPERIMENTAL, since it hasn't been michael@0: properly tested yet. michael@0: michael@0: - A Windows FNT specific API has been added, mostly to access font michael@0: headers. This is used by Wine. michael@0: michael@0: - TrueType tables without an `hmtx' table are now tolerated when michael@0: an incremental interface is used. This happens for certain michael@0: Type42 fonts passed from Ghostscript to FreeType. michael@0: michael@0: - The PFR font driver is now capable of returning the font family michael@0: and style names when they are available (instead of the sole michael@0: `FontID'). This is performed by parsing an *undocumented* michael@0: portion of the font file! michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - The path stroker in FT_STROKER_H has entered beta stage. It now michael@0: works very well, but its interface might change a bit in the michael@0: future. More on this in later releases. michael@0: michael@0: - The documentation for FT_Size_Metrics didn't appear properly in michael@0: the API reference. michael@0: michael@0: - The file docs/VERSION.DLL has been updated to explain versioning michael@0: with FreeType (i.e., comparing release/libtool/so numbers, and michael@0: how to use them in autoconf scripts). michael@0: michael@0: - The installation documentation has been seriously revamped. michael@0: Everything is now in the `docs' directory. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.1.3 and 2.1.2 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - FT_Vector_Transform had been incorrectly modified in 2.1.2, michael@0: resulting in incorrect transformations being applied (for michael@0: example, rotations were processed in opposite angles). michael@0: michael@0: - The format 8 and 12 TrueType charmap enumeration routines have michael@0: been fixed (FT_Get_Next_Char returned invalid values). michael@0: michael@0: - The PFR font driver returned incorrect advance widths if the michael@0: outline and metrics resolution defined in the font file were michael@0: different. michael@0: michael@0: - FT_Glyph_To_Bitmap now returns successfully when called with an michael@0: FT_BitmapGlyph argument (it previously returned an error). michael@0: michael@0: - A bug in the Type 1 loader that prevented valid font bounding michael@0: boxes to be loaded from multiple master fonts. michael@0: michael@0: - The SFNT validation code has been rewritten. FreeType can now michael@0: load `broken' fonts that were usable on Windows, but not with michael@0: previous versions of the library. michael@0: michael@0: - The computation of bearings in the BDF driver has been fixed. michael@0: michael@0: - The Postscript hinter crashed when trying to hint certain glyphs michael@0: (more precisely, when trying to apply hints to an empty glyph michael@0: outline). michael@0: michael@0: - The TrueType glyph loader now supports composites in `Apple michael@0: format' (they differ slightly from Microsoft/OpenType ones in michael@0: the way transformation offsets are computed). michael@0: michael@0: - FreeType was very slow at opening certain asian CID/CFF fonts, michael@0: due to fixed increment in dynamic array re-allocations. This michael@0: has been changed to exponential behaviour to get acceptable michael@0: performance. michael@0: michael@0: michael@0: michael@0: II. IMPORTANT CHANGES michael@0: michael@0: - The PCF driver now supports gzip-compressed font files natively. michael@0: This means that you will be able to use all these bitmap fonts michael@0: that come with XFree86 with FreeType (and libXft/libXft2, by michael@0: extension). michael@0: michael@0: - The automatic and postscript hinters have both been updated. michael@0: This results in a relatively important increase of rendering michael@0: quality since many nasty defaults have been suppressed. Please michael@0: visit the web page: michael@0: michael@0: http://www.freetype.org/hinting/smooth-hinting.html michael@0: michael@0: for additional details on this topic. michael@0: michael@0: - The `load_flags' parameter of `FT_Load_Glyph' is now an FT_Int32 michael@0: (instead of just being an FT_Int). This breaks source and michael@0: binary compatibility for 16bit systems only, while retaining michael@0: both of them for 32 and 64 bit ones. michael@0: michael@0: Some new flags have been added consequently: michael@0: michael@0: FT_LOAD_NO_AUTOHINT :: Disable the use of the auto-hinter michael@0: (but not native format hinters). michael@0: michael@0: FT_LOAD_TARGET_NORMAL :: Hint and render for normal michael@0: anti-aliased displays. michael@0: michael@0: FT_LOAD_TARGET_MONO :: Hint and render for 1-bit displays. michael@0: michael@0: FT_LOAD_TARGET_LCD :: Hint and render for horizontal RGB or michael@0: BGR sub-pixel displays (like LCD michael@0: screens). THIS IS STILL michael@0: EXPERIMENTAL! michael@0: michael@0: FT_LOAD_TARGET_LCD_V :: Same as FT_LOAD_TARGET_LCD, for michael@0: vertical sub-pixel displays (like michael@0: rotated LCD screens). THIS IS STILL michael@0: EXPERIMENTAL! michael@0: michael@0: FT_LOAD_MONOCHROME is still supported, but only affects michael@0: rendering, not the hinting. michael@0: michael@0: Note that the `ftview' demo program available in the `ft2demos' michael@0: package has been updated to support LCD-optimized display on michael@0: non-paletted displays (under Win32 and X11). michael@0: michael@0: - The PFR driver now supports embedded bitmaps (all formats michael@0: supported), and returns correct kerning metrics for all glyphs. michael@0: michael@0: - The TrueType charmap loader now supports certain `broken' fonts michael@0: that load under Windows without problems. michael@0: michael@0: - The cache API has been slightly modified (it's still a beta!): michael@0: michael@0: - The type FTC_ImageDesc has been removed; it is now replaced michael@0: by FTC_ImageTypeRec. Note that one of its fields is a michael@0: `load_flag' parameter for FT_Load_Glyph. michael@0: michael@0: - The field `num_grays' of FT_SBitRec has been changed to michael@0: `max_grays' in order to fit within a single byte. Its michael@0: maximum value is thus 255 (instead of 256 as previously). michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - Added support for the DESTDIR variable during `make install'. michael@0: This simplifies packaging of FreeType. michael@0: michael@0: - Included modified copies of the ZLib sources in `src/gzip' in michael@0: order to support gzip-compressed PCF fonts. We do not use the michael@0: system-provided zlib for now, though this is a probable michael@0: enhancement for future releases. michael@0: michael@0: - The DocMaker tool used to generate the on-line API reference has michael@0: been completely rewritten. It is now located in michael@0: `src/tools/docmaker/docmaker.py'. Features: michael@0: michael@0: - better cross-referenced output michael@0: - more polished output michael@0: - uses Python regular expressions (though it didn't speed the michael@0: program) michael@0: - much more modular structure, which allows for different michael@0: `backends' in order to generate HTML, XML, or whatever michael@0: format. michael@0: michael@0: One can regenerate the API reference by calling: michael@0: michael@0: python src/tools/docmaker/docmaker.py \ michael@0: --prefix=ft2 \ michael@0: --title=FreeType-2.1.3 \ michael@0: --output= michael@0: include/freetype/*.h \ michael@0: include/freetype/config/*.h \ michael@0: include/freetype/cache/*.h michael@0: michael@0: - A new, experimental, support for incremental font loading (i.e., michael@0: loading of fonts where the glyphs are not in the font file michael@0: itself, but provided by an external component, like a Postscript michael@0: interpreter) has been added by Graham Asher. This is still work michael@0: in progress, however. michael@0: michael@0: - A new, EXPERIMENTAL, path stroker has been added. It doesn't michael@0: suffer from severe rounding errors and treat bezier arcs michael@0: directly. Still work in progress (i.e. not part of the official michael@0: API). See the file for some of the michael@0: details. michael@0: michael@0: - The massive re-formatting of sources and internal re-design is michael@0: still under-way. Many internal functions, constants, and types michael@0: have been renamed. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.1.2 and 2.1.1 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Many font drivers didn't select a Unicode charmap by default michael@0: when a new face was opened (with the FT_CONFIG_OPTION_USE_CMAPS michael@0: options enabled), causing many applications to not be able to michael@0: display text correctly with the 2.1.x releases. michael@0: michael@0: - The PFR driver had a bug in its composite loading code that michael@0: produces incorrectly placed accents with many fonts. michael@0: michael@0: - The Type42 driver crashed sometimes due to a nasty bug. michael@0: michael@0: - The Type 1 custom encoding charmap didn't handle the case where michael@0: the first glyph index wasn't 0. michael@0: michael@0: - A serious typo in the TrueType composite loader produced michael@0: incorrectly placed glyphs in fonts like `Wingdings' and a few michael@0: others. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - The Win32 Visual C++ project file has been updated to include michael@0: the PFR driver as well. michael@0: michael@0: - `freetype.m4' is now installed by default by `make install' on michael@0: Unix systems. michael@0: michael@0: - The function FT_Get_PS_Font_Info now works with CID and Type42 michael@0: fonts as well. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.1.1 and 2.1.0 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - The `version_info' returned by `freetype-config' in 2.1.0 michael@0: returned an invalid value. It now returns 9:1:3 (2.0.9 returned michael@0: 9:0:3). michael@0: michael@0: - Version 2.1.0 couldn't be linked against applications on Win32 michael@0: and Amiga systems due to a new debug function that wasn't michael@0: properly propagated to the system-specific directory in michael@0: `builds'. michael@0: michael@0: - Various MacOS and Mac OS X specific fixes. michael@0: michael@0: - Fixed a bug in the TrueType charmap validation routines that michael@0: made version 2.1.0 too restrictive -- many popular fonts have michael@0: been rejected. michael@0: michael@0: - There was still a very small difference between the monochrome michael@0: glyph bitmaps produced by FreeType 1.x and FreeType 2.x with the michael@0: bytecode interpreter enabled. This was caused by an invalid michael@0: flag setting in the TrueType glyph loader, making the rasterizer michael@0: change its drop-out control mode. Now the results should michael@0: _really_ be completely identical. michael@0: michael@0: - The TrueType name table loader has been improved to support many michael@0: popular though buggy Asian fonts. It now ignores empty name michael@0: entries, invalid pointer offsets and a few other incorrect michael@0: subtleties. Moreover, name strings are now loaded on demand, michael@0: which reduces the memory load of many faces (e.g. the ARIAL.TTF michael@0: font file contains a 10kByte name table with 70 names). michael@0: michael@0: - Fixed a bug in the Postscript hinter that prevented family blues michael@0: substitution to happen correctly. michael@0: michael@0: michael@0: II. NEW FEATURES michael@0: michael@0: - Three new font drivers in this release: michael@0: michael@0: * A BDF font driver, contributed by Franco Zappa Nardelli, michael@0: heavily modified by Werner Lemberg. It also supports michael@0: anti-aliased bitmaps (using a slightly extended BDF format). michael@0: michael@0: * A Type42 font driver, contributed by Roberto Alameda. It is michael@0: still experimental but seems to work relatively well. michael@0: michael@0: * A PFR font driver, contributed by David Turner himself. It michael@0: doesn't support PFR hinting -- note that BitStream has at michael@0: least two patents on this format! michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - The cache sub-system has been optimized in important ways. michael@0: Cache hits are now significantly faster. For example, using the michael@0: CMap cache is about twice faster than calling FT_Get_Char_Index michael@0: on most platforms. Similarly, using an SBit cache is about five michael@0: times faster than loading the bitmaps from a bitmap file, and michael@0: 300 to 500 times faster than generating them from a scalable michael@0: format. michael@0: michael@0: Note that you should recompile your sources if you designed a michael@0: custom cache class for the FT2 Cache subsystem, since the michael@0: changes performed are source, but not binary, compatible. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.1.0 and 2.0.9 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - The TrueType bytecode interpreter has been fixed to produce michael@0: _exactly_ the same output as FreeType 1.x. Previous differences michael@0: were due to slightly distinct fixed-point computation routines michael@0: used to perform dot products and vector length measurements. michael@0: michael@0: It seems that native TrueType hinting is _extremely_ sensitive michael@0: to rounding errors. The required vector computation routines michael@0: have been optimized and placed within the `ttinterp.c' file. michael@0: michael@0: - Fixed the parsing of accelerator tables in the PCF font driver. michael@0: michael@0: - Fixed the Type1 glyph loader routine used to compute the font's michael@0: maximum advance width. michael@0: michael@0: michael@0: II. NEW FEATURES michael@0: michael@0: - The `configure' script used on Unix systems has been modified to michael@0: check that GNU Make is being used to build the library. michael@0: Otherwise, it will display a message proposing to use the michael@0: GNUMAKE environment variable to name it. michael@0: michael@0: The Unix-specific file README.UNX has been modified accordingly. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - The FreeType License in `docs/FTL.TXT' has been updated to michael@0: include a proposed preferred disclaimer. If you are using michael@0: FreeType in your products, you are encouraged (but not mandated) michael@0: to use the following text in your documentation: michael@0: michael@0: """ michael@0: Portions of this software are copyright © 1996-2002 The michael@0: FreeType Project (www.freetype.org). All rights reserved. michael@0: """ michael@0: michael@0: - The default size of the render pool has been reduced to 16kByte. michael@0: This shouldn't result in any noticeable performance penalty, michael@0: unless you are using the engine as-is to render very large and michael@0: complex glyphs. michael@0: michael@0: - The FreeType 2 redesign has begun. More information can be michael@0: found at this URL: michael@0: michael@0: http://www.freetype.org/freetype2/redesign.html michael@0: michael@0: The following internal changes have been performed within the michael@0: sources of this release: michael@0: michael@0: - Many internal types have been renamed to increase michael@0: consistency. The following should be true, except for michael@0: public types: michael@0: michael@0: * All structure types have a name ending in `Rec' (short michael@0: for `record'). michael@0: michael@0: * A pointer-to-structure type has the same name as the michael@0: structure, _without_ the `Rec' suffix. michael@0: michael@0: Example: michael@0: michael@0: typedef struct FooRec_ michael@0: { michael@0: ... michael@0: michael@0: } FooRec, *Foo; michael@0: michael@0: - Many internal macros have been renamed to increase michael@0: consistency. The following should be true: michael@0: michael@0: * All macros have a name beginning with `FT_'. This michael@0: required a few changes like michael@0: michael@0: ALLOC => FT_ALLOC michael@0: FREE => FT_FREE michael@0: REALLOC => FT_REALLOC michael@0: michael@0: * All macros are completely UPPERCASE. This required a michael@0: few changes like: michael@0: michael@0: READ_Short => FT_READ_SHORT michael@0: NEXT_Short => FT_NEXT_SHORT michael@0: GET_ULongLE => FT_GET_ULONG_LE michael@0: MEM_Set => FT_MEM_SET michael@0: MEM_Copy => FT_MEM_COPY michael@0: etc. michael@0: michael@0: * Whenever possible, all macro names follow the michael@0: FT__ pattern. For example michael@0: michael@0: ACCESS_Frame => FT_FRAME_ENTER michael@0: FORGET_Frame => FT_FRAME_EXIT michael@0: EXTRACT_Frame => FT_FRAME_EXTRACT michael@0: RELEASE_Frame => FT_FRAME_RELEASE michael@0: michael@0: FILE_Pos => FT_STREAM_POS michael@0: FILE_Seek => FT_STREAM_SEEK michael@0: FILE_Read => FT_STREAM_READ michael@0: FILE_ReadAt => FT_STREAM_READ_AT michael@0: READ_Fields => FT_STREAM_READ_FIELDS michael@0: michael@0: - Many internal functions have been renamed to follow the michael@0: FT__ pattern. For example: michael@0: michael@0: FT_Seek_Stream => FT_Stream_Seek michael@0: FT_Read_Stream_At => FT_Stream_ReadAt michael@0: FT_Done_Stream => FT_Stream_Close michael@0: FT_New_Stream => FT_Stream_Open michael@0: FT_New_Memory_Stream => FT_Stream_OpenMemory michael@0: FT_Extract_Frame => FT_Stream_ExtractFrame michael@0: michael@0: Note that method names do not contain `_'. michael@0: michael@0: - The FT_ALLOC_ARRAY and FT_REALLOC_ARRAY have been replaced michael@0: with FT_NEW_ARRAY and FT_RENEW_ARRAY which do not take a michael@0: type as the fourth argument. Instead, the array element michael@0: type size is computed automatically from the type of the michael@0: target pointer used. michael@0: michael@0: - A new object class, FT_CMap, has been introduced. These michael@0: internal objects are used to model character maps. This michael@0: eases the support of additional charmap types within the michael@0: engine. michael@0: michael@0: - A new configuration file named `ftstdlib.h' has been added michael@0: to `include/freetype/config'. It is used to define aliases michael@0: for _every_ routine of the ISO C library that the font michael@0: engine uses. Each aliases has a `ft_' prefix michael@0: (e.g. `ft_strlen' is an alias for `strlen'). michael@0: michael@0: This is used to ease the porting of FreeType 2 to exotic michael@0: runtime environments where the ISO C Library isn't available michael@0: (e.g. XFree86 extension modules). michael@0: michael@0: More details are available in the `ChangeLog' file. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.0.9 and 2.0.8 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Certain fonts like `foxjump.ttf' contain broken name tables with michael@0: invalid entries and wild offsets. This caused FreeType to crash michael@0: when trying to load them. michael@0: michael@0: The SFNT `name' table loader has been fixed to be able to michael@0: support these strange fonts. michael@0: michael@0: Moreover, the code in charge of processing this table has been michael@0: changed to always favour Windows-formatted entries over other michael@0: ones. Hence, a font that works on Windows but not on the Mac michael@0: will load cleanly in FreeType and report accurate values for michael@0: Family & PostScript names. michael@0: michael@0: - The CID font driver has been fixed. It unfortunately returned a michael@0: Postscript Font name with a leading slash, as in michael@0: `/MunhwaGothic-Regular'. michael@0: michael@0: - FreeType 2 should now compile fine on AIX 4.3.3 as a shared michael@0: library. michael@0: michael@0: - A bug in the Postscript hinter has been found and fixed, michael@0: removing un-even stem widths at small pixel sizes (like 14-17). michael@0: michael@0: This improves the quality of a certain number of Postscript michael@0: fonts. michael@0: michael@0: michael@0: II. NEW FEATURES michael@0: michael@0: - A new function named `FT_Library_Version' has been added to michael@0: return the current library's major, minor, and patch version michael@0: numbers. This is important since the macros FREETYPE_MAJOR, michael@0: FREETYPE_MINOR, and FREETYPE_PATCH cannot be used when the michael@0: library is dynamically linked by a program. michael@0: michael@0: - Two new APIs have been added: `FT_Get_First_Char' and michael@0: `FT_Get_Next_Char'. michael@0: michael@0: Together, these can be used to iterate efficiently over the michael@0: currently selected charmap of a given face. Read the API michael@0: reference for more details. michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - The FreeType sources are under heavy internal re-factoring. As michael@0: a consequence, we have created a branch named `STABLE' on the michael@0: CVS to hold all future releases/fixes in the 2.0.x family. michael@0: michael@0: The HEAD branch now contains the re-factored sources and michael@0: shouldn't be used for testing or packaging new releases. In michael@0: case you would like to access the 2.0.9 sources from our CVS michael@0: repository, use the tag `VER-2-0-9'. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.0.8 and 2.0.7 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - There was a small but nasty bug in `freetype-config.in' which michael@0: caused the `freetype-config' script to fail on Unix. michael@0: michael@0: This didn't prevent the installation of the library or even its michael@0: execution, but caused problems when trying to compile many Unix michael@0: packages that depend on it. michael@0: michael@0: - Some TrueType or OpenType fonts embedded in PDF documents do not michael@0: have a 'cmap', 'post' and 'name' as is required by the michael@0: specification. FreeType no longer refuses to load such fonts. michael@0: michael@0: - Various fixes to the PCF font driver. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.0.7 and 2.0.6 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - Fixed two bugs in the Type 1 font driver. The first one michael@0: resulted in a memory leak in subtle cases. The other one caused michael@0: FreeType to crash when trying to load `.gsf' files (Ghostscript michael@0: so-called Postscript fonts). michael@0: michael@0: (This made _many_ KDE applications crash on certain systems. michael@0: FreeType _is_ becoming a critical system component on Linux :-) michael@0: michael@0: - Fixed a memory leak in the CFF font driver. michael@0: michael@0: - Fixed a memory leak in the PCF font driver. michael@0: michael@0: - Fixed the Visual C++ project file michael@0: `builds/win32/visualc/freetype.dsp' since it didn't include the michael@0: Postscript hinter component, causing errors at build time. michael@0: michael@0: - Fixed a small rendering bug in the anti-aliased renderer that michael@0: only occurred when trying to draw thin (less than 1 pixel) michael@0: strokes. michael@0: michael@0: - Fixed `builds/unix/freetype2.a4' which is used to generate a michael@0: valid `freetype2.m4' for use with autoconf. michael@0: michael@0: - Fixed the OpenVMS Makefiles. michael@0: michael@0: michael@0: II. MISCELLANEOUS michael@0: michael@0: - Added `configure' and `install' scripts to the top-level michael@0: directory. A GNU-style installation is thus now easily possible michael@0: with michael@0: michael@0: ./configure michael@0: make michael@0: make install michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.0.6 and 2.0.5 michael@0: michael@0: I. IMPORTANT BUG FIXES michael@0: michael@0: - It wasn't possible to load embedded bitmaps when the auto-hinter michael@0: was used. This is now fixed. michael@0: michael@0: - The TrueType font driver didn't load some composites properly michael@0: (the sub-glyphs were slightly shifted, and this was only michael@0: noticeable when using monochrome rendering). michael@0: michael@0: - Various fixes to the auto-hinter. They merely improve the michael@0: output of sans-serif fonts. Note that there are still problems michael@0: with serifed fonts and composites (accented characters). michael@0: michael@0: - All scalable font drivers erroneously returned un-fitted glyph michael@0: advances when hinting was requested. This created problems for michael@0: a number of layout applications. This is a very old bug that michael@0: got undetected mainly because most test/demo program perform michael@0: rounding explicitly or implicitly (through the cache). michael@0: michael@0: - `FT_Glyph_To_Bitmap' did erroneously modify the source glyph in michael@0: certain cases. michael@0: michael@0: - `glnames.py' still contained a bug that made FreeType return michael@0: invalid names for certain glyphs. michael@0: michael@0: - The library crashed when loading certain Type 1 fonts like michael@0: `sadn.pfb' (`Stalingrad Normal'), which appear to contain michael@0: pathetic font info dictionaries. michael@0: michael@0: - The TrueType glyph loader is now much more paranoid and checks michael@0: everything when loading a given glyph image. This was necessary michael@0: to avoid problems (crashes and/or memory overwrites) with broken michael@0: fonts that came from a really buggy automatic font converter. michael@0: michael@0: michael@0: II. IMPORTANT UPDATES AND NEW FEATURES michael@0: michael@0: - Important updates to the Mac-specific parts of the library. michael@0: michael@0: - The caching sub-system has been completely re-designed, and its michael@0: API has evolved (the old one is still supported for backward michael@0: compatibility). michael@0: michael@0: The documentation for it is not yet completed, sorry. For now, michael@0: you are encouraged to continue using the old API. However, the michael@0: ftview demo program in the ft2demos package has already been michael@0: updated to use the new caching functions. michael@0: michael@0: - A new charmap cache is provided too. See `FTC_CMapCache'. This michael@0: is useful to perform character code -> glyph index translations michael@0: quickly, without the need for an opened FT_Face. michael@0: michael@0: - A NEW POSTSCRIPT HINTER module has been added to support native michael@0: hints in the following formats: PostScript Type 1, PostScript michael@0: CID, and CFF/CEF. michael@0: michael@0: Please test! Note that the auto-hinter produces better results michael@0: for a number of badly-hinted fonts (mostly auto-generated ones) michael@0: though. michael@0: michael@0: - A memory debugger is now part of the standard FreeType sources. michael@0: To enable it, define FT_DEBUG_MEMORY in michael@0: , and recompile the library. michael@0: michael@0: Additionally, define the _environment_ variable FT_DEBUG_MEMORY michael@0: and run any program using FreeType. When the library is exited, michael@0: a summary of memory footprints and possible leaks will be michael@0: displayed. michael@0: michael@0: This works transparently with _any_ program that uses FreeType. michael@0: However, you will need a lot of memory to use this (allocated michael@0: blocks are never released to the heap to detect double deletes michael@0: easily). michael@0: michael@0: michael@0: III. MISCELLANEOUS michael@0: michael@0: - We are aware of subtle differences between the output of michael@0: FreeType versions 1 and 2 when it comes to monochrome michael@0: TrueType-hinted glyphs. These are most probably due to small michael@0: differences in the monochrome rasterizers and will be worked out michael@0: in an upcoming release. michael@0: michael@0: - We have decided to fork the sources in a `stable' branch, and an michael@0: `unstable' one, since FreeType is becoming a critical component michael@0: of many Unix systems. michael@0: michael@0: The next bug-fix releases of the library will be named 2.0.7, michael@0: 2.0.8, etc., while the `2.1' branch will contain a version of michael@0: the sources where we will start major reworking of the library's michael@0: internals, in order to produce FreeType 2.2.0 (or even 3.0) in a michael@0: more distant future. michael@0: michael@0: We also hope that this scheme will allow much more frequent michael@0: releases than in the past. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.0.5 and 2.0.4 michael@0: michael@0: NOTE THAT 2.0.5 DOES NOT CONTAIN THE POSTSCRIPT HINTER. THIS MODULE michael@0: WILL BE PART OF THE NEXT RELEASE (EITHER 2.0.6 or 2.1) michael@0: michael@0: - Fixed a bug that made certain glyphs, like `Cacute', `cacute' and michael@0: `lslash' unavailable from Unicode charmaps of Postscript fonts. michael@0: This prevented the correct display of Polish text, for example. michael@0: michael@0: - The kerning table of Type 1 fonts was loaded by FreeType, when its michael@0: AFM file was attached to its face, but the michael@0: FT_FACE_FLAG_HAS_KERNING bit flags was not set correctly, michael@0: preventing FT_Get_Kerning to return meaningful values. michael@0: michael@0: - Improved SFNT (TrueType & OpenType) charmap support. Slightly michael@0: better performance, as well as support for the new formats defined michael@0: by the OpenType 1.3 specification (8, 10, and 12) michael@0: michael@0: - Fixed a serious typo in `src/base/ftcalc.c' which caused invalid michael@0: computations in certain rare cases, producing ugly artefacts. michael@0: michael@0: - The size of the EM square is computed with a more accurate michael@0: algorithm for Postscript fonts. The old one caused slight errors michael@0: with embedded fonts found in PDF documents. michael@0: michael@0: - Fixed a bug in the cache manager that prevented normal LRU michael@0: behaviour within the cache manager, causing unnecessary reloads michael@0: (for FT_Face and FT_Size objects only). michael@0: michael@0: - Added a new function named `FT_Get_Name_Index' to retrieve the michael@0: glyph index of a given glyph name, when found in a face. michael@0: michael@0: - Added a new function named `FT_Get_Postscript_Name' to retrieve michael@0: the `unique' Postscript font name of a given face. michael@0: michael@0: - Added a new public header size named FT_SIZES_H (or michael@0: ) providing new FT_Size-management functions: michael@0: FT_New_Size, FT_Activate_Size, FT_Done_Size. michael@0: michael@0: - Fixed a reallocation bug that generated a dangling pointer (and michael@0: possibly memory leaks) with Postscript fonts (in michael@0: src/psaux/psobjs.c). michael@0: michael@0: - Many fixes for 16-bit correctness. michael@0: michael@0: - Removed many pedantic compiler warnings from the sources. michael@0: michael@0: - Added an Amiga build directory in `builds/amiga'. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.0.4 and 2.0.3 michael@0: michael@0: - Fixed a rather annoying bug that was introduced in 2.0.3. Namely, michael@0: the font transformation set through FT_Set_Transform was applied michael@0: twice to auto-hinted glyphs, resulting in incorrectly rotated text michael@0: output. michael@0: michael@0: - Fixed _many_ compiler warnings. FT2 should now compile cleanly michael@0: with Visual C++'s most pedantic warning level (/W4). It already michael@0: compiled fine with GCC and a few other compilers. michael@0: michael@0: - Fixed a bug that prevented the linear advance width of composite michael@0: TrueType glyphs to be correctly returned. michael@0: michael@0: - Fixed the Visual C++ project files located in michael@0: `builds/win32/visualc' (previous versions used older names of the michael@0: library). michael@0: michael@0: - Many 32-bit constants have an `L' appended to their value, in michael@0: order to improve the 16-bitness of the code. Someone is actually michael@0: trying to use FT2 on an Atari ST machine! michael@0: michael@0: - Updated the `builds/detect.mk' file in order to automatically michael@0: build FT2 on AIX systems. AIX uses `/usr/sbin/init' instead of michael@0: `/sbin/init' and wasn't previously detected as a Unix platform by michael@0: the FreeType build system. michael@0: michael@0: - Updated the Unix-specific portions of the build system (new michael@0: libtool version, etc.). michael@0: michael@0: - The SFNT kerning loader now ensures that the table is sorted michael@0: (since some problem fonts do not meet this requirement). michael@0: michael@0: michael@0: ======================================================================= michael@0: michael@0: CHANGES BETWEEN 2.0.3 and 2.0.2 michael@0: michael@0: I. CHANGES TO THE MODULES / FONT DRIVERS michael@0: michael@0: - THE AUTO-HINTER HAS BEEN SLIGHTLY IMPROVED, in order to fix michael@0: several annoying artefacts, mainly: michael@0: michael@0: - Blue zone alignment of horizontal stems wasn't performed michael@0: correctly, resulting in artefacts like the `d' being placed michael@0: one pixel below the `b' in some fonts like Time New Roman. michael@0: michael@0: - Overshoot thresholding wasn't performed correctly, creating michael@0: unpleasant artefacts at large character pixel sizes. michael@0: michael@0: - Composite glyph loading has been simplified. This gets rid michael@0: of various artefacts where the components of a composite michael@0: glyphs were not correctly spaced. michael@0: michael@0: These are the last changes to the current auto-hinting module. michael@0: A new hinting sub-system is currently in the work in order to michael@0: support native hints in Type 1 / CFF / OpenType fonts, as well michael@0: as globally improve rendering. michael@0: michael@0: - The PCF driver has been fixed. It reported invalid glyph michael@0: dimensions for the fonts available on Solaris. michael@0: michael@0: - The Type 1, CID and CFF drivers have been modified to fix the michael@0: computation of the EM size. michael@0: michael@0: - The Type 1 driver has been fixed to avoid a dangerous bug that michael@0: crashed the library with non-conforming fonts (i.e. ones that do michael@0: not place the .notdef glyph at position 0). michael@0: michael@0: - The TrueType driver had a rather subtle bug (dangling pointer michael@0: when loading composite glyphs) that could crash the library in michael@0: rare occasions! michael@0: michael@0: michael@0: II. HIGH-LEVEL API CHANGES michael@0: michael@0: - The error code enumeration values have been changed. An error michael@0: value is decomposed in a generic error code, and a module michael@0: number. see for details. michael@0: michael@0: - A new public header file has been introduced, named michael@0: FT_TRIGONOMETRY_H (include/freetype/fttrigon.h), providing michael@0: trigonometric functions to compute sines, cosines, arctangents, michael@0: etc. with 16.16 fixed precision. The implementation is based on michael@0: the CORDIC algorithm and is very fast while being sufficiently michael@0: accurate. michael@0: michael@0: michael@0: III. INTERNALS michael@0: michael@0: - Added BeOS-specific files in the old build sub-system. Note michael@0: that no changes were required to compile the library with Jam. michael@0: michael@0: - The configuration is now capable of automatically detecting michael@0: 64-bit integers on a set of predefined compilers (GCC, Visual michael@0: C++, Borland C++) and will use them by default. This provides a michael@0: small performance boost. michael@0: michael@0: - A small memory leak that happened when opening 0-sized files michael@0: (duh!) have been fixed. michael@0: michael@0: - Fixed bezier stack depth bug in the routines provided by the michael@0: FT_BBOX_H header file. Also fixed similar bugs in the michael@0: rasterizers. michael@0: michael@0: - The outline bounding box code has been rewritten to use direct michael@0: computations, instead of bezier sub-division, to compute the michael@0: exact bounding box of glyphs. This is slightly slower but more michael@0: accurate. michael@0: michael@0: - The build system has been improved and fixed, mainly to support michael@0: `make' on Windows 2000 correctly, avoid problems with `make michael@0: distclean' on non Unix systems, etc. michael@0: michael@0: - Hexadecimal constants have been suffixed with `U' to avoid michael@0: problems with certain compilers on 64-bit platforms. michael@0: michael@0: - A new directory named `src/tools' has been created. It contains michael@0: Python scripts and simple unit test programs used to develop the michael@0: library. michael@0: michael@0: - The DocMaker tool has been moved from `docs' to `src/tools' and michael@0: has been updated with the following: michael@0: michael@0: - Now accepts the `--title=XXXX' or `-t XXXX' option from the michael@0: command line to set the project's name in the generated API michael@0: reference. michael@0: michael@0: - Now accepts the `--output=DIR' or `-o DIR' option from the michael@0: command line to set the output directory for all generated michael@0: HTML files. michael@0: michael@0: - Now accepts the `--prefix=XXXX' or `-p XXX' option from the michael@0: command line to set the file prefix to use for all michael@0: generated HTML files. michael@0: michael@0: - Now generates the current time/data on each generated page michael@0: in order to distinguish between versions. michael@0: michael@0: DocMaker can be used with other projects now, not only FT2 michael@0: (e.g. MLib, FTLayout, etc.). michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.0.2 and 2.0.1 michael@0: michael@0: I. CHANGES TO THE MODULES / FONT DRIVERS michael@0: michael@0: - THE TRUETYPE BYTECODE INTERPRETER IS NOW TURNED OFF, in order to michael@0: avoid legal problems with the Apple patents. It seems that we michael@0: mistakenly turned this option on in previous releases of the michael@0: build. michael@0: michael@0: Note that if you want to use the bytecode interpreter in order michael@0: to get high-quality TrueType rendering, you will need to toggle michael@0: by hand the definition of the michael@0: TT_CONFIG_OPTION_BYTECODE_INTERPRETER macro in the file michael@0: `include/freetype/config/ftoption.h'. michael@0: michael@0: - The CFF driver has been improved by Tom Kacvinsky and Sander van michael@0: der Wal: michael@0: michael@0: * Support for `seac' emulation. michael@0: * Support for `dotsection'. michael@0: * Support for retrieving glyph names through michael@0: `FT_Get_Glyph_Name'. michael@0: michael@0: The first two items are necessary to correctly a large number of michael@0: Type 1 fonts converted to the CFF formats by Adobe Acrobat. michael@0: michael@0: - The Type 1 driver was also improved by Tom & others: michael@0: michael@0: * Better EM size computation. michael@0: * Better support for synthetic (transformed) fonts. michael@0: * The Type 1 driver returns the charstrings corresponding to michael@0: each glyph in the `glyph->control_data' field after a call to michael@0: `FT_Load_Glyph' (thanks Ha Shao). michael@0: michael@0: - Various other bugfixes, including the following: michael@0: michael@0: * Fixed a nasty memory leak in the Type 1 driver. michael@0: * The autohinter and the pcf driver used static writable data michael@0: when they shouldn't. michael@0: * Many casts were added to make the code more 64-bits safe. It michael@0: also now compiles on Windows XP 64-bits without warnings. michael@0: * Some incorrect writable statics were removed in the `autohint' michael@0: and `pcf' drivers. FreeType 2 now compiles on Epoc again. michael@0: michael@0: michael@0: II. CHANGES TO THE HIGH-LEVEL API michael@0: michael@0: - The library header files inclusion scheme has been changed. The michael@0: old scheme looked like: michael@0: michael@0: #include michael@0: #include michael@0: #include michael@0: #include michael@0: michael@0: Now you should use: michael@0: michael@0: #include michael@0: #include FT_FREETYPE_H michael@0: #include FT_GLYPH_H michael@0: #include FT_CACHE_H michael@0: #include FT_CACHE_IMAGE_H michael@0: michael@0: NOTE THAT THE OLD INCLUSION SCHEME WILL STILL WORK WITH THIS michael@0: RELEASE. HOWEVER, WE DO NOT GUARANTEE THAT THIS WILL STILL BE michael@0: TRUE IN THE NEXT ONE (A.K.A. FREETYPE 2.1). michael@0: michael@0: The file is used to define the header filename michael@0: macros. The complete and commented list of macros is available michael@0: in the API reference under the section name `Header File Macros' michael@0: in Chapter I. michael@0: michael@0: For more information, see section I of the following document: michael@0: michael@0: http://www.freetype.org/ michael@0: freetype2/docs/tutorial/step1.html michael@0: michael@0: or michael@0: michael@0: http://freetype.sourceforge.net/ michael@0: freetype2/docs/tutorial/step1.html michael@0: michael@0: - Many, many comments have been added to the public source file in michael@0: order to automatically generate the API Reference through the michael@0: `docmaker.py' Python script. michael@0: michael@0: The latter has been updated to support the grouping of sections michael@0: in chapters and better index sort. See: michael@0: michael@0: http://www.freetype.org/freetype2/docs/reference/ft2-toc.html michael@0: michael@0: michael@0: III. CHANGES TO THE BUILD PROCESS michael@0: michael@0: - If you are not building FreeType 2 with its own build system michael@0: (but with your own Makefiles or project files), you will need to michael@0: be aware that the build process has changed a little bit. michael@0: michael@0: You don't need to put the `src' directory in the include path michael@0: when compiling any FT2 component. Instead, simply put the michael@0: component's directory in the current include path. michael@0: michael@0: So, if you were doing something like: michael@0: michael@0: cc -c -Iinclude -Isrc src/base/ftbase.c michael@0: michael@0: change the line to: michael@0: michael@0: cc -c -Iinclude -Isrc/base src/base/ftbase.c michael@0: michael@0: If you were doing something like: michael@0: michael@0: cd src/base michael@0: cc -c -I../../include -I.. ftbase.c michael@0: michael@0: change it to: michael@0: michael@0: cd src/base michael@0: cc -c -I../../include ftbase.c michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN 2.0.1 and 2.0 michael@0: michael@0: 2.0.1 introduces a few changes: michael@0: michael@0: - Fixed many bugs related to the support of CFF / OpenType fonts. michael@0: These formats are now much better supported though there is michael@0: still work planned to deal with charset tables and PDF-embedded michael@0: CFF files that use the old `seac' command. michael@0: michael@0: - The library could not be compiled in debug mode with a very michael@0: small number of C compilers whose pre-processors didn't michael@0: implement the `##' directive correctly (i.e. per se the ANSI C michael@0: specification!) An elegant fix was found. michael@0: michael@0: - Added support for the free Borland command-line C++ Builder michael@0: compiler. Use `make setup bcc32'. Also fixed a few source michael@0: lines that generated new warnings with BCC32. michael@0: michael@0: - Fixed a bug in FT_Outline_Get_BBox when computing the extrema of michael@0: a conic Bezier arc. michael@0: michael@0: - Updated the INSTALL file to add IDE compilation. michael@0: michael@0: - Other minor bug fixes, from invalid Type 1 style flags to michael@0: correct support of synthetic (obliqued) fonts in the michael@0: auto-hinter, better support for embedded bitmaps in a SFNT font. michael@0: michael@0: - Fixed some problems with `freetype-config'. michael@0: michael@0: Finally, the `standard' scheme for including FreeType headers is now michael@0: gradually changing, but this will be explained in a later release michael@0: (probably 2.0.2). michael@0: michael@0: And very special thanks to Tom Kacvinsky and YAMANO-UCHI Hidetoshi michael@0: for their contributions! michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: CHANGES BETWEEN beta8 and 2.0 michael@0: michael@0: - Changed the default installation path for public headers from michael@0: `include/freetype' to `include/freetype2'. michael@0: michael@0: Also added a new `freetype-config' that is automatically generated michael@0: and installed on Unix and Cygwin systems. The script itself is michael@0: used to retrieve the current install path, C compilation flags as michael@0: well as linker flags. michael@0: michael@0: - Fixed several small bugs: michael@0: michael@0: * Incorrect max advance width for fixed-pitch Type 1 fonts. michael@0: * Incorrect glyph names for certain TrueType fonts. michael@0: * The glyph advance was not copied when FT_Glyph_To_Bitmap was michael@0: called. michael@0: * The linearHoriAdvance and linearVertAdvance fields were not michael@0: correctly returned for glyphs processed by the auto-hinter. michael@0: * `type1z' renamed back to `type1'; the old `type1' module has michael@0: been removed. michael@0: michael@0: - Revamped the build system to make it a lot more generic. This michael@0: will allow us to re-use nearly un-modified in lots of other michael@0: projects (including FreeType Layout). michael@0: michael@0: - Changed `cid' to use `psaux' too. michael@0: michael@0: - Added the cache sub-system. See as well as michael@0: the sources in `src/cache'. Note that it compiles but is still michael@0: untested for now. michael@0: michael@0: - Updated `docs/docmaker.py', a draft API reference is available at michael@0: http://www.freetype.org/ft2api.html. michael@0: michael@0: - Changed `type1' to use `psaux'. michael@0: michael@0: - Created a new module named `psaux' to hold the Type 1 & Type 2 michael@0: parsing routines. It should be used by `type1', `cid', and `cff' michael@0: in the future. michael@0: michael@0: - Fixed an important bug in `FT_Glyph_Get_CBox'. michael@0: michael@0: - Fixed some compiler warnings that happened since the TrueType michael@0: bytecode decoder was deactivated by default. michael@0: michael@0: - Fixed two memory leaks: michael@0: michael@0: * The memory manager (16 bytes) isn't released in michael@0: FT_Done_FreeType! michael@0: * Using custom input streams, the copy of the original stream was michael@0: never released. michael@0: michael@0: - Fixed the auto-hinter by performing automatic computation of the michael@0: `filling direction' of each glyph. This is done through a simple michael@0: and fast approximation, and seems to work (problems spotted by michael@0: Werner though). The Arphic fonts are a lot nicer though there are michael@0: still a lot of things to do to handle Asian fonts correctly. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: BETA-8 (RELEASE CANDIDATE) CHANGES michael@0: michael@0: - Deactivated the TrueType bytecode interpreter by default. michael@0: michael@0: - Deactivated the `src/type1' font driver. Now `src/type1z' is used michael@0: by default. michael@0: michael@0: - Updates to the build system. We now compile the library correctly michael@0: under Unix system through `configure' which is automatically michael@0: called on the first `make' invocation. michael@0: michael@0: - Added the auto-hinting module! Fixing some bugs here and there. michael@0: michael@0: - Found some bugs in the composite loader (seac) of the Type1-based michael@0: font drivers. michael@0: michael@0: - Renamed the directory `freetype2/config' to `freetype2/builds' and michael@0: updated all relevant files. michael@0: michael@0: - Found a memory leak in the `type1' driver. michael@0: michael@0: - Incorporated Tom's patches to support flex operators correctly in michael@0: OpenType/CFF fonts. Now all I need is to support pure CFF and CEF michael@0: fonts to be done with this driver :-) michael@0: michael@0: - Added the Windows FNT/FON driver in `src/winfonts'. For now, it michael@0: always `simulates' a Unicode charmap, so it shouldn't be michael@0: considered completed right now. michael@0: michael@0: It is there to be more a proof of concept than anything else michael@0: anyway. The driver is a single C source file, that compiles to 3 michael@0: Kb of code. michael@0: michael@0: I'm still working on the PCF/BDF drivers, but I'm too lazy to michael@0: finish them now. michael@0: michael@0: - CHANGES TO THE HIGH-LEVEL API michael@0: michael@0: * FT_Get_Kerning has a new parameter that allows you to select the michael@0: coordinates of the kerning vector (font units, scaled, scaled + michael@0: grid-fitted). michael@0: * The outline functions are now in and not michael@0: part of anymore. michael@0: * now contains declarations for michael@0: FT_New_Library, FT_Done_Library, FT_Add_Default_Modules. michael@0: * The so-called convenience functions have moved from `ftoutln.c' michael@0: to `ftglyph.c', and are thus available with this optional michael@0: component of the library. They are declared in michael@0: now. michael@0: * Anti-aliased rendering is now the default for FT_Render_Glyph michael@0: (i.e. corresponds to render_mode == 0 == ft_render_mode_normal). michael@0: To generate a monochrome bitmap, use ft_render_mode_mono, or the michael@0: FT_LOAD_MONOCHROME flag in FT_Load_Glyph/FT_Load_Char. michael@0: FT_LOAD_ANTI_ALIAS is still defined, but values to 0. michael@0: * now include , michael@0: solving a few headaches :-) michael@0: * The type FT_GlyphSlotRec has now a `library' field. michael@0: michael@0: - CHANGES TO THE `ftglyph.h' API michael@0: michael@0: This API has been severely modified in order to make it simpler, michael@0: clearer, and more efficient. It certainly now looks like a real michael@0: `glyph factory' object, and allows client applications to manage michael@0: (i.e. transform, bbox and render) glyph images without ever michael@0: knowing their original format. michael@0: michael@0: - Added support for CID-keyed fonts to the CFF driver. Maybe michael@0: support for pure CFF + CEF fonts should come in? michael@0: michael@0: - Cleaned up source code in order to avoid two functions with the michael@0: same name. Also changed the names of the files in `type1z' from michael@0: `t1XXXX' to `z1XXXX' in order to avoid any conflicts. michael@0: michael@0: `make multi' now works well :-) michael@0: michael@0: Also removed the use of `cidafm' for now, even if the source files michael@0: are still there. This functionality will certainly go into a michael@0: specific module. michael@0: michael@0: - ADDED SUPPORT FOR THE AUTO-HINTER michael@0: michael@0: It works :-) I have a demo program which simply is a copy of michael@0: `ftview' that does a `FT_Add_Module(library, michael@0: &autohinter_module_class)' after library initialization, and Type michael@0: 1 & OpenType/CFF fonts are now hinted. michael@0: michael@0: CID fonts are not hinted, as they include no charmap and the michael@0: auto-hinter doesn't include `generic' global metrics computations michael@0: yet. michael@0: michael@0: Now, I need to release this thing to the FreeType 2 source. michael@0: michael@0: - CHANGES TO THE RENDERER MODULES michael@0: michael@0: The monochrome and smooth renderers are now in two distinct michael@0: directories, namely `src/raster1' and `src/smooth'. Note that the michael@0: old `src/renderer' is now gone. michael@0: michael@0: I ditched the 5-gray-levels renderers. Basically, it involved a michael@0: simple #define toggle in 'src/raster1/ftraster.c'. michael@0: michael@0: FT_Render_Glyph, FT_Outline_Render & FT_Outline_Get_Bitmap now michael@0: select the best renderer available, depending on render mode. If michael@0: the current renderer for a given glyph image format isn't capable michael@0: of supporting the render mode, another one will be found in the michael@0: library's list. This means that client applications do not need michael@0: to switch or set the renderers themselves (as in the latest michael@0: change), they'll get what they want automatically. At last. michael@0: michael@0: Changed the demo programs accordingly. michael@0: michael@0: - MAJOR INTERNAL REDESIGN: michael@0: michael@0: A lot of internal modifications have been performed lately on the michael@0: source in order to provide the following enhancements: michael@0: michael@0: * More generic module support: michael@0: michael@0: The FT_Module type is now defined to represent a handle to a michael@0: given module. The file contains the michael@0: FT_Module_Class definition, as well as the module-loading public michael@0: API. michael@0: michael@0: The FT_Driver type is still defined, and still represents a michael@0: pointer to a font driver. Note that FT_Add_Driver is replaced michael@0: by FT_Add_Module, FT_Get_Driver by FT_Get_Module, etc. michael@0: michael@0: * Support for generic glyph image types: michael@0: michael@0: The FT_Renderer type is a pointer to a module used to perform michael@0: various operations on glyph image. michael@0: michael@0: Each renderer is capable of handling images in a single format michael@0: (e.g. ft_glyph_format_outline). Its functions are used to: michael@0: michael@0: - transform an glyph image michael@0: - render a glyph image into a bitmap michael@0: - return the control box (dimensions) of a given glyph image michael@0: michael@0: The scan converters `ftraster.c' and `ftgrays.c' have been moved michael@0: to the new directory `src/renderer', and are used to provide two michael@0: default renderer modules. michael@0: michael@0: One corresponds to the `standard' scan-converter, the other to michael@0: the `smooth' one. michael@0: michael@0: he current renderer can be set through the new function michael@0: FT_Set_Renderer. michael@0: michael@0: The old raster-related function FT_Set_Raster, FT_Get_Raster and michael@0: FT_Set_Raster_Mode have now disappeared, in favor of the new: michael@0: michael@0: FT_Get_Renderer michael@0: FT_Set_Renderer michael@0: michael@0: See the file for more details. michael@0: michael@0: These changes were necessary to properly support different michael@0: scalable formats in the future, like bi-color glyphs, etc. michael@0: michael@0: * Glyph loader object: michael@0: michael@0: A new internal object, called a 'glyph loader' has been michael@0: introduced in the base layer. It is used by all scalable format michael@0: font drivers to load glyphs and composites. michael@0: michael@0: This object has been created to reduce the code size of each michael@0: driver, as each one of them basically re-implemented its michael@0: functionality. michael@0: michael@0: See and the FT_GlyphLoader type for michael@0: more information. michael@0: michael@0: * FT_GlyphSlot has new fields: michael@0: michael@0: In order to support extended features (see below), the michael@0: FT_GlyphSlot structure has a few new fields: michael@0: michael@0: linearHoriAdvance: michael@0: michael@0: This field gives the linearly scaled (i.e. scaled but michael@0: unhinted) advance width for the glyph, expressed as a 16.16 michael@0: fixed pixel value. This is useful to perform WYSIWYG text. michael@0: michael@0: linearVertAdvance: michael@0: This field gives the linearly scaled advance height for the michael@0: glyph (relevant in vertical glyph layouts only). This is michael@0: useful to perform WYSIWYG text. michael@0: michael@0: Note that the two above field replace the removed `metrics2' michael@0: field in the glyph slot. michael@0: michael@0: advance: michael@0: This field is a vector that gives the transformed advance for michael@0: the glyph. By default, it corresponds to the advance width, michael@0: unless FT_LOAD_VERTICAL_LAYOUT was specified when calling michael@0: FT_Load_Glyph or FT_Load_Char. michael@0: michael@0: bitmap_left: michael@0: This field gives the distance in integer pixels from the michael@0: current pen position to the left-most pixel of a glyph image michael@0: IF IT IS A BITMAP. It is only valid when the `format' field michael@0: is set to `ft_glyph_format_bitmap', for example, after calling michael@0: the new function FT_Render_Glyph. michael@0: michael@0: bitmap_top: michael@0: This field gives the distance in integer pixels from the michael@0: current pen position (located on the baseline) to the top-most michael@0: pixel of the glyph image IF IT IS A BITMAP. Positive values michael@0: correspond to upwards Y. michael@0: michael@0: loader: michael@0: This is a new private field for the glyph slot. Client michael@0: applications should not touch it. michael@0: michael@0: michael@0: * Support for transforms and direct rendering in FT_Load_Glyph: michael@0: michael@0: Most of the functionality found in has been michael@0: moved to the core library. Hence, the following: michael@0: michael@0: - A transform can be specified for a face through michael@0: FT_Set_Transform. this transform is applied by FT_Load_Glyph michael@0: to scalable glyph images (i.e. NOT TO BITMAPS) before the michael@0: function returns, unless the bit flag FT_LOAD_IGNORE_TRANSFORM michael@0: was set in the load flags. michael@0: michael@0: - Once a glyph image has been loaded, it can be directly michael@0: converted to a bitmap by using the new FT_Render_Glyph michael@0: function. Note that this function takes the glyph image from michael@0: the glyph slot, and converts it to a bitmap whose properties michael@0: are returned in `face.glyph.bitmap', `face.glyph.bitmap_left' michael@0: and `face.glyph.bitmap_top'. The original native image might michael@0: be lost after the conversion. michael@0: michael@0: - When using the new bit flag FT_LOAD_RENDER, the FT_Load_Glyph michael@0: and FT_Load_Char functions will call FT_Render_Glyph michael@0: automatically when needed. michael@0: michael@0: - Reformatted all modules source code in order to get rid of the michael@0: basic data types redifinitions (i.e. `TT_Int' instead of `FT_Int', michael@0: `T1_Fixed' instead of `FT_Fixed'). Hence the format-specific michael@0: prefixes like `TT_', `T1_', `T2_' and `CID_' are only used for michael@0: relevant structures. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: OLD CHANGES FOR BETA 7 michael@0: michael@0: - bug-fixed the OpenType/CFF parser. It now loads and displays my michael@0: two fonts nicely, but I'm pretty certain that more testing is michael@0: needed :-) michael@0: michael@0: - fixed the crummy Type 1 hinter, it now handles accented characters michael@0: correctly (well, the accent is not always well placed, but that's michael@0: another problem..) michael@0: michael@0: - added the CID-keyed Type 1 driver in `src/cid'. Works pretty well michael@0: for only 13 Kb of code ;-) Doesn't read AFM files though, nor the michael@0: really useful CMAP files.. michael@0: michael@0: - fixed two bugs in the smooth renderer (src/base/ftgrays.c). michael@0: Thanks to Boris Letocha for spotting them and providing a fix. michael@0: michael@0: - fixed potential `divide by zero' bugs in ftcalc.c. michael@0: michael@0: - added source code for the OpenType/CFF driver (still incomplete michael@0: though..) michael@0: michael@0: - modified the SFNT driver slightly to perform more robust header michael@0: checks in TT_Load_SFNT_Header. This prevents certain font files michael@0: (e.g. some Type 1 Multiple Masters) from being incorrectly michael@0: `recognized' as TrueType font files.. michael@0: michael@0: - moved a lot of stuff from the TrueType driver to the SFNT module, michael@0: this allows greater code re-use between font drivers michael@0: (e.g. TrueType, OpenType, Compact-TrueType, etc..) michael@0: michael@0: - added a tiny segment cache to the SFNT Charmap 4 decoder, in order michael@0: to minimally speed it up.. michael@0: michael@0: - added support for Multiple Master fonts in `type1z'. There is michael@0: also a new file named which defines functions to michael@0: manage them from client applications. michael@0: michael@0: The new file `src/base/ftmm.c' is also optional to the engine.. michael@0: michael@0: - various formatting changes (e.g. EXPORT_DEF -> FT_EXPORT_DEF) + michael@0: small bug fixes in FT_Load_Glyph, the `type1' driver, etc.. michael@0: michael@0: - a minor fix to the Type 1 driver to let them apply the font matrix michael@0: correctly (used for many oblique fonts..) michael@0: michael@0: - some fixes for 64-bit systems (mainly changing some FT_TRACE calls michael@0: to use %p instead of %lx). Thanks to Karl Robillard. michael@0: michael@0: - fixed some bugs in the sbit loader (src/base/sfnt/ttsbit.c) + michael@0: added a new flag, FT_LOAD_CROP_BITMAP to query that bitmaps be michael@0: cropped when loaded from a file (maybe I should move the bitmap michael@0: cropper to the base layer ??). michael@0: michael@0: - changed the default number of gray levels of the smooth renderer michael@0: to 256 (instead of the previous 128). Of course, the human eye michael@0: can't see any difference ;-) michael@0: michael@0: - removed TT_MAX_SUBGLYPHS, there is no static limit on the number michael@0: of subglyphs in a TrueType font now.. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: OLD CHANGES 16 May 2000 michael@0: michael@0: - tagged `BETA-6' in the CVS tree. This one is a serious release michael@0: candidate even though it doesn't incorporate the auto-hinter yet.. michael@0: michael@0: - various obsolete files were removed, and copyright header updated michael@0: michael@0: - finally updated the standard raster to fix the monochrome michael@0: rendering bug + re-enable support for 5-gray levels anti-aliasing michael@0: (suck, suck..) michael@0: michael@0: - created new header files, and modified sources accordingly: michael@0: michael@0: michael@0: - simple FreeType types, without the API michael@0: michael@0: - definition of memory-management macros michael@0: michael@0: - added the `DSIG' (OpenType Digital Signature) tag to michael@0: michael@0: michael@0: - light update/cleaning of the build system + changes to the sources michael@0: in order to get rid of _all_ compiler warnings with three michael@0: compilers, i.e: michael@0: michael@0: gcc with `-ansi -pedantic -Wall -W', Visual C++ with `/W3 /WX' and michael@0: LCC michael@0: michael@0: IMPORTANT NOTE FOR WIN32-LCC USERS: michael@0: | michael@0: | It seems the C pre-processor that comes with LCC is broken, it michael@0: | doesn't recognize the ANSI standard directives # and ## michael@0: | correctly when one of the argument is a macro. Also, michael@0: | something like: michael@0: | michael@0: | #define F(x) print##x michael@0: | michael@0: | F(("hello")) michael@0: | michael@0: | will get incorrectly translated to: michael@0: | michael@0: | print "hello") michael@0: | michael@0: | by its pre-processor. For this reason, you simply cannot build michael@0: | FreeType 2 in debug mode with this compiler.. michael@0: michael@0: - yet another massive grunt work. I've changed the definition of michael@0: the EXPORT_DEF, EXPORT_FUNC, BASE_DEF & BASE_FUNC macros. These michael@0: now take an argument, which is the function's return value type. michael@0: michael@0: This is necessary to compile FreeType as a DLL on Windows and michael@0: OS/2. Depending on the compiler used, a compiler-specific keyword michael@0: like __export or __system must be placed before (VisualC++) or michael@0: after (BorlandC++) the type.. michael@0: michael@0: Of course, this needed a lot of changes throughout the source code michael@0: to make it compile again... All cleaned up now, apparently.. michael@0: michael@0: Note also that there is a new EXPORT_VAR macro defined to allow michael@0: the _declaration_ of an exportable public (constant) michael@0: variable. This is the case of the raster interfaces (see michael@0: ftraster.h and ftgrays.h), as well as each module's interface (see michael@0: sfdriver.h, psdriver.h, etc..) michael@0: michael@0: - new feature: it is now possible to pass extra parameters to font michael@0: drivers when creating a new face object. For now, michael@0: this capability is unused. It could however prove to michael@0: be useful in a near future.. michael@0: michael@0: the FT_Open_Args structure was changes, as well as the internal michael@0: driver interface (the specific `init_face' module function has michael@0: now a different signature). michael@0: michael@0: - updated the tutorial (not finished though). michael@0: michael@0: - updated the top-level BUILD document michael@0: michael@0: - fixed a potential memory leak that could occur when loading michael@0: embedded bitmaps. michael@0: michael@0: - added the declaration of FT_New_Memory_Face in michael@0: , as it was missing from the public header michael@0: (the implementation was already in `ftobjs.c'). michael@0: michael@0: - the file has been seriously updated in order michael@0: to allow the automatic generation of error message tables. See michael@0: the comments within it for more information. michael@0: michael@0: - major directory hierarchy re-organisation. This was done for two michael@0: things: michael@0: michael@0: * first, to ease the `manual' compilation of the library by michael@0: requiring at lot less include paths :-) michael@0: michael@0: * second, to allow external programs to effectively access michael@0: internal data fields. For example, this can be extremely michael@0: useful if someone wants to write a font producer or a font michael@0: manager on top of FreeType. michael@0: michael@0: Basically, you should now use the 'freetype/' prefix for header michael@0: inclusion, as in: michael@0: michael@0: #include michael@0: #include michael@0: michael@0: Some new include sub-directories are available: michael@0: michael@0: a. the `freetype/config' directory, contains two files used to michael@0: configure the build of the library. Client applications michael@0: should not need to look at these normally, but they can if michael@0: they want. michael@0: michael@0: #include michael@0: #include michael@0: michael@0: b. the `freetype/internal' directory, contains header files that michael@0: describes library internals. These are the header files that michael@0: were previously found in the `src/base' and `src/shared' michael@0: directories. michael@0: michael@0: michael@0: As usual, the build system and the demos have been updated to michael@0: reflect the change.. michael@0: michael@0: Here's a layout of the new directory hierarchy: michael@0: michael@0: TOP_DIR michael@0: include/ michael@0: freetype/ michael@0: freetype.h michael@0: ... michael@0: config/ michael@0: ftoption.h michael@0: ftconfig.h michael@0: ftmodule.h michael@0: michael@0: internal/ michael@0: ftobjs.h michael@0: ftstream.h michael@0: ftcalc.h michael@0: ... michael@0: michael@0: src/ michael@0: base/ michael@0: ... michael@0: michael@0: sfnt/ michael@0: psnames/ michael@0: truetype/ michael@0: type1/ michael@0: type1z/ michael@0: michael@0: michael@0: Compiling a module is now much easier, for example, the following michael@0: should work when in the TOP_DIR directory on an ANSI build: michael@0: michael@0: gcc -c -I./include -I./src/base src/base/ftbase.c michael@0: gcc -c -I./include -I./src/sfnt src/sfnt/sfnt.c michael@0: etc.. michael@0: michael@0: (of course, using -Iconfig/ if you provide system-specific michael@0: configuration files). michael@0: michael@0: - updated the structure of FT_Outline_Funcs in order to allow direct michael@0: coordinate scaling within the outline decomposition routine (this michael@0: is important for virtual `on' points with TrueType outlines) + michael@0: updates to the rasters to support this.. michael@0: michael@0: - updated the OS/2 table loading code in `src/sfnt/ttload.c' in michael@0: order to support version 2 of the table (see OpenType 1.2 spec) michael@0: michael@0: - created `include/tttables.h' and `include/t1tables.h' to allow michael@0: client applications to access some of the SFNT and T1 tables of a michael@0: face with a procedural interface (see `FT_Get_Sfnt_Table') + michael@0: updates to internal source files to reflect the change.. michael@0: michael@0: - some cleanups in the source code to get rid of warnings when michael@0: compiling with the `-Wall -W -ansi -pedantic' options in gcc. michael@0: michael@0: - debugged and moved the smooth renderer to `src/base/ftgrays.c' and michael@0: its header to `include/ftgrays.h' michael@0: michael@0: - updated TT_MAX_SUBGLYPHS to 96 as some CJK fonts have composites michael@0: with up to 80 sub-glyphs !! Thanks to Werner michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: OLD CHANGES - 14-apr-2000 michael@0: michael@0: - fixed a bug in the TrueType glyph loader that prevented the michael@0: correct loading of some CJK glyphs in mingli.ttf michael@0: michael@0: - improved the standard Type 1 hinter in `src/type1' michael@0: michael@0: - fixed two bugs in the experimental Type 1 driver in `src/type1z' michael@0: to handle the new XFree86 4.0 fonts (and a few other ones..) michael@0: michael@0: - the smooth renderer is now complete and supports sub-banding to michael@0: render large glyphs at high speed. However, it is still located michael@0: in `demos/src/ftgrays.c' and should move to the library itself in michael@0: the next beta. NOTE: The smooth renderer doesn't compile in michael@0: stand-alone mode anymore, but this should be fixed RSN.. michael@0: michael@0: - introduced convenience functions to more easily deal with glyph michael@0: images, see `include/ftglyph.h' for more details, as well as the michael@0: new demo program named `demos/src/ftstring.c' that demonstrates michael@0: its use michael@0: michael@0: - implemented FT_LOAD_NO_RECURSE in both the TrueType and Type 1 michael@0: drivers (this is required by the auto-hinter to improve its michael@0: results). michael@0: michael@0: - changed the raster interface, in order to allow client michael@0: applications to provide their own span-drawing callbacks. michael@0: However, only the smooth renderer supports this. See michael@0: `FT_Raster_Params' in the file `include/ftimage.h'. michael@0: michael@0: - fixed a small bug in FT_MulFix that caused incorrect transform michael@0: computation! michael@0: michael@0: - Note: The tutorial is out-of-date. michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: OLD CHANGES - 12-mar-2000 michael@0: michael@0: - changed the layout of configuration files : now, all ANSI michael@0: configuration files are located in michael@0: `freetype2/config'. System-specific over-rides can be placed in michael@0: `freetype2/config/'. michael@0: michael@0: - moved all configuration macros to `config/ftoption.h' michael@0: michael@0: - improvements in the Type 1 driver with AFM support michael@0: michael@0: - changed the fields in the FT_Outline structure : the old `flags' michael@0: array is re-named `tags', while all ancient flags are encoded into michael@0: a single unsigned int named `flags'. michael@0: michael@0: - introduced new flags in FT_Outline.flags (see michael@0: ft_outline_.... enums in `ftimage.h'). michael@0: michael@0: - changed outline functions to `FT_Outline_' syntax michael@0: michael@0: - added a smooth anti-alias renderer to the demonstration programs michael@0: michael@0: - added Mac graphics driver (thanks Just) michael@0: michael@0: - FT_Open_Face changed in order to received a pointer to a michael@0: FT_Open_Args descriptor.. michael@0: michael@0: - various cleanups, a few more API functions implemented (see michael@0: FT_Attach_File) michael@0: michael@0: - updated some docs michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: OLD CHANGES - 22-feb-2000 michael@0: michael@0: - introduced the `psnames' module. It is used to: michael@0: michael@0: o convert a Postscript glyph name into the equivalent Unicode michael@0: character code (used by the Type 1 driver(s) to synthesize on michael@0: the fly a Unicode charmap). michael@0: michael@0: o provide an interface to retrieve the Postscript names of the michael@0: Macintosh, Adobe Standard & Adobe Expert character codes. michael@0: (the Macintosh names are used by the SFNT-module postscript michael@0: names support routines, while the other two tables are used by michael@0: the Type 1 driver(s)). michael@0: michael@0: - introduced the `type1z' alternate Type 1 driver. This is a (still michael@0: experimental) driver for the Type 1 format that will ultimately michael@0: replace the one in `src/type1'. It uses pattern matching to load michael@0: data from the font, instead of a finite state analyzer. It works michael@0: much better than the `old' driver with `broken' fonts. It is also michael@0: much smaller (under 15 Kb). michael@0: michael@0: - the Type 1 drivers (both in `src/type1' and `src/type1z') are michael@0: nearly complete. They both provide automatic Unicode charmap michael@0: synthesis through the `psnames' module. No re-encoding vector is michael@0: needed. (note that they still leak memory due to some code michael@0: missing, and I'm getting lazy). michael@0: michael@0: Trivial AFM support has been added to read kerning information but michael@0: wasn't exactly tested as it should ;-) michael@0: michael@0: - The TrueType glyph loader has been seriously rewritten (see the michael@0: file `src/truetype/ttgload.c'. It is now much, much simpler as michael@0: well as easier to read, maintain and understand :-) Preliminary michael@0: versions introduced a memory leak that has been reported by Jack michael@0: Davis, and is now fixed.. michael@0: michael@0: - introduced the new `ft_glyph_format_plotter', used to represent michael@0: stroked outlines like Windows `Vector' fonts, and certain Type 1 michael@0: fonts like `Hershey'. The corresponding raster will be written michael@0: soon. michael@0: michael@0: - FT_New_Memory_Face is gone. Likewise, FT_Open_Face has a new michael@0: interface that uses a structure to describe the input stream, the michael@0: driver (if required), etc.. michael@0: michael@0: michael@0: TODO michael@0: michael@0: - Write FT_Get_Glyph_Bitmap and FT_Load_Glyph_Bitmap michael@0: michael@0: - Add a function like FT_Load_Character(face, char_code, load_flags) michael@0: that would really embed a call to FT_Get_Char_Index then michael@0: FT_Load_Glyph to ease developer's work. michael@0: michael@0: - Update the tutorial! michael@0: michael@0: - consider adding support for Multiple Master fonts in the Type 1 michael@0: drivers. michael@0: michael@0: - Test the AFM routines of the Type 1 drivers to check that kerning michael@0: information is returned correctly. michael@0: michael@0: - write a decent auto-gridding component !! We need this to release michael@0: FreeType 2.0 gold ! michael@0: michael@0: michael@0: less urgent needs: michael@0: michael@0: - add a CFF/Type2 driver michael@0: - add a BDF driver michael@0: - add a FNT/PCF/HBF driver michael@0: - add a Speedo driver from the X11 sources michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: OLDER CHANGES - 27-jan-2000 michael@0: michael@0: - updated the `sfnt' module interface to allow several SFNT-based michael@0: drivers to co-exist peacefully michael@0: michael@0: - updated the `T1_Face' type to better separate Postscript font michael@0: content from the rest of the FT_Face structure. Might be used michael@0: later by the CFF/Type2 driver.. michael@0: michael@0: - added an experimental replacement Type 1 driver featuring advanced michael@0: (and speedy) pattern matching to retrieve the data from postscript michael@0: fonts. michael@0: michael@0: - very minor changes in the implementation of FT_Set_Char_Size and michael@0: FT_Set_Pixel_Sizes (they now implement default to lighten the font michael@0: driver's code). michael@0: michael@0: michael@0: ====================================================================== michael@0: michael@0: OLD MESSAGE michael@0: michael@0: This file summarizes the changes that occurred since the last `beta' michael@0: of FreeType 2. Because the list is important, it has been divided into michael@0: separate sections: michael@0: michael@0: Table Of Contents: michael@0: michael@0: I High-Level Interface (easier !) michael@0: II Directory Structure michael@0: III Glyph Image Formats michael@0: IV Build System michael@0: V Portability michael@0: VI Font Drivers michael@0: michael@0: michael@0: ---------------------------------------------------------------------- michael@0: michael@0: High-Level Interface: michael@0: michael@0: The high-level API has been considerably simplified. Here is how: michael@0: michael@0: - resource objects have disappeared. this means that face objects michael@0: can now be created with a single function call (see FT_New_Face michael@0: and FT_Open_Face) michael@0: michael@0: - when calling either FT_New_Face & FT_Open_Face, a size object michael@0: and a glyph slot object are automatically created for the face, michael@0: and can be accessed through `face->glyph' and `face->size' if michael@0: one really needs to. In most cases, there's no need to call michael@0: FT_New_Size or FT_New_Glyph. michael@0: michael@0: - similarly, FT_Load_Glyph now only takes a `face' argument michael@0: (instead of a glyph slot and a size). Also, its `result' michael@0: parameter is gone, as the glyph image type is returned in the michael@0: field `face->glyph.format' michael@0: michael@0: - the list of available charmaps is directly accessible through michael@0: `face->charmaps', counting `face->num_charmaps' elements. Each michael@0: charmap has an 'encoding' field which specifies which known michael@0: encoding it deals with. Valid values are, for example: michael@0: michael@0: ft_encoding_unicode (for ASCII, Latin-1 and Unicode) michael@0: ft_encoding_apple_roman michael@0: ft_encoding_sjis michael@0: ft_encoding_adobe_standard michael@0: ft_encoding_adobe_expert michael@0: michael@0: other values may be added in the future. Each charmap still michael@0: holds its `platform_id' and `encoding_id' values in case the michael@0: encoding is too exotic for the current library michael@0: michael@0: michael@0: ---------------------------------------------------------------------- michael@0: michael@0: Directory Structure: michael@0: michael@0: Should seem obvious to most of you: michael@0: michael@0: freetype/ michael@0: config/ -- configuration sub-makefiles michael@0: ansi/ michael@0: unix/ -- platform-specific configuration files michael@0: win32/ michael@0: os2/ michael@0: msdos/ michael@0: michael@0: include/ -- public header files, those to be included michael@0: directly by client apps michael@0: michael@0: src/ -- sources of the library michael@0: base/ -- the base layer michael@0: sfnt/ -- the sfnt `driver' (see the drivers section michael@0: below) michael@0: truetype/ -- the truetype driver michael@0: type1/ -- the type1 driver michael@0: shared/ -- some header files shared between drivers michael@0: michael@0: demos/ -- demos/tools michael@0: michael@0: docs/ -- documentation (a bit empty for now) michael@0: michael@0: michael@0: ---------------------------------------------------------------------- michael@0: michael@0: Glyph Image Formats: michael@0: michael@0: Drivers are now able to register new glyph image formats within the michael@0: library. For now, the base layer supports of course bitmaps and michael@0: vector outlines, but one could imagine something different like michael@0: colored bitmaps, bi-color vectors or whatever else (Metafonts anyone michael@0: ??). michael@0: michael@0: See the file `include/ftimage.h'. Note also that the type michael@0: FT_Raster_Map is gone, and is now replaced by FT_Bitmap, which michael@0: should encompass all known bitmap types. michael@0: michael@0: Each new image format must provide at least one `raster', i.e. a michael@0: module capable of transforming the glyph image into a bitmap. It's michael@0: also possible to change the default raster used for a given glyph michael@0: image format. michael@0: michael@0: The default outline scan-converter now uses 128 levels of grays by michael@0: default, which tends to smooth many things. Note that the demo michael@0: programs have been updated significantly in order to display these.. michael@0: michael@0: michael@0: ---------------------------------------------------------------------- michael@0: michael@0: Build system: michael@0: michael@0: You still need GNU Make to build the library. The build system has michael@0: been very seriously re-vamped in order to provide things like : michael@0: michael@0: - automatic host platform detection (reverting to 'config/ansi' if michael@0: it is not detected, with pseudo-standard compilation flags) michael@0: michael@0: - the ability to compile from the Makefiles with very different and michael@0: exotic compilers. Note that linking the library can be difficult michael@0: for some platforms. michael@0: michael@0: For example, the file `config/win32/lcclib.bat' is invoked by the michael@0: build system to create the `.lib' file with LCC-Win32 because its michael@0: librarian has too many flaws to be invoked directly from the michael@0: Makefile. michael@0: michael@0: Here's how it works: michael@0: michael@0: - the first time you type `make', the build system runs a series of michael@0: sub-makefiles in order to detect your host platform. It then michael@0: dumps what it found, and creates a file called `config.mk' in the michael@0: current directory. This is a sub-Makefile used to define many michael@0: important Make variables used to build the library. michael@0: michael@0: - the second time, the build system detects the `config.mk' then use michael@0: it to build the library. All object files go into 'obj' by michael@0: default, as well as the library file, but this can easily be michael@0: changed. michael@0: michael@0: Note that you can run `make setup' to force another host platform michael@0: detection even if a `config.mk' is present in the current michael@0: directory. Another solution is simply to delete the file, then michael@0: re-run make. michael@0: michael@0: Finally, the default compiler for all platforms is gcc (for now, michael@0: this will hopefully changed in the future). You can however specify michael@0: a different compiler by specifying it after the 'setup' target as michael@0: in: michael@0: michael@0: gnumake setup lcc on Win32 to use the LCC compiler michael@0: gnumake setup visualc on Win32 to use Visual C++ michael@0: michael@0: See the file `config//detect.mk' for a list of supported michael@0: compilers for your platforms. michael@0: michael@0: It should be relatively easy to write new detection rules files and michael@0: config.mk.. michael@0: michael@0: Finally, to build the demo programs, go to `demos' and launch GNU michael@0: Make, it will use the `config.mk' in the top directory to build the michael@0: test programs.. michael@0: michael@0: michael@0: ---------------------------------------------------------------------- michael@0: michael@0: Portability: michael@0: michael@0: In the previous beta, a single FT_System object was used to michael@0: encompass all low-level operations like thread synchronisation, michael@0: memory management and i/o access. This has been greatly simplified: michael@0: michael@0: - thread synchronisation has been dropped, for the simple reason michael@0: that the library is already re-entrant, and that if you really michael@0: need two threads accessing the same FT_Library, you should michael@0: really synchronize access to it yourself with a simple mutex. michael@0: michael@0: - memory management is performed through a very simple object michael@0: called `FT_Memory', which really is a table containing a table michael@0: of pointers to functions like malloc, realloc and free as well michael@0: as some user data (closure). michael@0: michael@0: - resources have disappeared (they created more problems than they michael@0: solved), and i/o management have been simplified greatly as a michael@0: result. Streams are defined through FT_Stream objects, which michael@0: can be either memory-based or disk-based. michael@0: michael@0: Note that each face has its own stream, which is closed only michael@0: when the face object is destroyed. Hence, a function like michael@0: TT_Flush_Face in 1.x cannot be directly supported. However, if michael@0: you really need something like this, you can easily tailor your michael@0: own streams to achieve the same feature at a lower level (and michael@0: use FT_Open_Face instead of FT_New_Face to create the face). michael@0: michael@0: See the file `include/ftsystem.h' for more details, as well as the michael@0: implementations found in `config/unix' and `config/ansi'. michael@0: michael@0: michael@0: ---------------------------------------------------------------------- michael@0: michael@0: Font Drivers: michael@0: michael@0: The Font Driver interface has been modified in order to support michael@0: extensions & versioning. michael@0: michael@0: michael@0: The list of the font drivers that are statically linked to the michael@0: library at compile time is managed through a new configuration file michael@0: called `config//ftmodule.h'. michael@0: michael@0: This file is autogenerated when invoking `make modules'. This michael@0: target will parse all sub-directories of 'src', looking for a michael@0: `module.mk' rules file, used to describe the driver to the build michael@0: system. michael@0: michael@0: Hence, one should call `make modules' each time a font driver is michael@0: added or removed from the `src' directory. michael@0: michael@0: Finally, this version provides a `pseudo-driver' in `src/sfnt'. michael@0: This driver doesn't support font files directly, but provides michael@0: services used by all TrueType-like font drivers. Hence, its code is michael@0: shared between the TrueType & OpenType font formats, and possibly michael@0: more formats to come if we're lucky.. michael@0: michael@0: michael@0: ---------------------------------------------------------------------- michael@0: michael@0: Extensions support: michael@0: michael@0: The extensions support is inspired by the one found in 1.x. michael@0: michael@0: Now, each font driver has its own `extension registry', which lists michael@0: which extensions are available for the font faces managed by the michael@0: driver. michael@0: michael@0: Extension ids are now strings, rather than 4-byte tags, as this is michael@0: usually more readable. michael@0: michael@0: Each extension has: michael@0: - some data, associated to each face object michael@0: - an interface (table of function pointers) michael@0: michael@0: An extension that is format-specific should simply register itself michael@0: to the correct font driver. Here is some example code: michael@0: michael@0: // Registering an extensions michael@0: // michael@0: FT_Error FT_Init_XXXX_Extension( FT_Library library ) michael@0: { michael@0: FT_DriverInterface* tt_driver; michael@0: michael@0: driver = FT_Get_Driver( library, "truetype" ); michael@0: if (!driver) return FT_Err_Unimplemented_Feature; michael@0: michael@0: return FT_Register_Extension( driver, &extension_class ); michael@0: } michael@0: michael@0: michael@0: // Implementing the extensions michael@0: // michael@0: FT_Error FT_Proceed_Extension_XXX( FT_Face face ) michael@0: { michael@0: FT_XXX_Extension ext; michael@0: FT_XXX_Extension_Interface ext_interface; michael@0: michael@0: ext = FT_Get_Extension( face, "extensionid", &ext_interface ); michael@0: if (!ext) return error; michael@0: michael@0: return ext_interface->do_it(ext); michael@0: } michael@0: michael@0: ------------------------------------------------------------------------ michael@0: michael@0: Copyright 2000-2013 by michael@0: David Turner, Robert Wilhelm, and Werner Lemberg. michael@0: michael@0: This file is part of the FreeType project, and may only be used, michael@0: modified, and distributed under the terms of the FreeType project michael@0: license, LICENSE.TXT. By continuing to use, modify, or distribute this michael@0: file you indicate that you have read the license and understand and michael@0: accept it fully. michael@0: michael@0: michael@0: Local Variables: michael@0: version-control: never michael@0: coding: utf-8 michael@0: End: michael@0: michael@0: --- end of CHANGES ---