|
1 Mozilla Build Verification Scripts |
|
2 --- |
|
3 |
|
4 -- |
|
5 Contents |
|
6 -- |
|
7 |
|
8 updates -> AUS and update verification |
|
9 l10n -> l10n vs. en-US verification |
|
10 common -> useful utility scripts |
|
11 |
|
12 -- |
|
13 Update verification |
|
14 -- |
|
15 |
|
16 verify.sh |
|
17 does a low-level check of all advertised MAR files. Expects to have a |
|
18 file named all-locales, but does not (yet) handle platform exceptions, so |
|
19 these should be removed from the locales file. |
|
20 |
|
21 prints errors on both STDOUT and STDIN, the intention is to run the |
|
22 script with STDOUT redirected to an output log. If there is not output |
|
23 on the console and an exit code of 0 then all tests pass; otherwise one |
|
24 or more tests failed. |
|
25 |
|
26 Does the following: |
|
27 |
|
28 1) download update.xml from AUS for a particular release |
|
29 2) download the partial and full mar advertised |
|
30 3) check that the partial and full match the advertised size and sha1sum |
|
31 4) downloads the latest release, and an older release |
|
32 5) applies MAR to the older release, and compares the two releases. |
|
33 |
|
34 Step 5 is repeated for both the complete and partial MAR. |
|
35 |
|
36 Expects to have an updates.cfg file, describing all releases to try updating |
|
37 from. |
|
38 |
|
39 - |
|
40 Valid Platforms for AUS |
|
41 - |
|
42 Linux_x86-gcc3 |
|
43 Darwin_Universal-gcc3 |
|
44 Linux_x86-gcc3 |
|
45 WINNT_x86-msvc |
|
46 Darwin_ppc-gcc3 |
|
47 |
|
48 -- |
|
49 l10n verification |
|
50 -- |
|
51 |
|
52 verify_l10n.sh |
|
53 unpacks an en-US build for a particular release/platform, and |
|
54 then unpacks and compares all locales for that particular release/platform. |
|
55 |
|
56 Expects to have a file named all-locales, but does not (yet) handle platform |
|
57 exceptions, so these should be removed from the locales file. |
|
58 |
|
59 Best practice is to take a directory full of diffs for a particular release |
|
60 and compare to a directory full of diffs for the current release, to see |
|
61 what l10n changes have occurred. For maintenance releases, this should |
|
62 be slim to none. |
|
63 |