Preparing patch now
Actually ... Bug 1224528 added this: https://hg.mozilla.org/mozilla-central/rev/735f81d9fd96c9d19d3f2fdfda29a14c1a066d22#l1.12 ; [Default Preferences] ; All the pref files must be part of base to prevent migration bugs @RESPATH@/browser/@PREF_DIR@/firefox.js @RESPATH@/browser/@PREF_DIR@/firefox-branding.js @RESPATH@/greprefs.js @RESPATH@/defaults/autoconfig/prefcalls.js @RESPATH@/browser/defaults/permissions +@RESPATH@/browser/defaults/blocklists Is this only applicable to Firefox? SM has @RESPATH@/defaults/permissions but TB/IM have not. So perhaps better to disable the test?
Created attachment 8840306 [details] [diff] [review] 1341983-skip-test_blocklist_certificates.patch Patch to disable test.
Comment on attachment 8840306 [details] [diff] [review] 1341983-skip-test_blocklist_certificates.patch Mathieu, perhaps you can approve this half-line change or advise us otherwise. We're currently seeing test failures all over our tree which I'd like to resolve.
Comment on attachment 8840306 [details] [diff] [review] 1341983-skip-test_blocklist_certificates.patch Disabling this test is one way to go. The other way would be to ship the JSON files on Thunderbird too (see services/blocklists/moz.build). Note: In the long term (Bug 1257565) the JSON diff-based system is going to replace the blocklist XML daily download.
(In reply to Mathieu Leplatre (:leplatrem) from comment #4) > Disabling this test is one way to go. The other way would be to ship the > JSON files on Thunderbird too (see services/blocklists/moz.build). Can you please advise us on what is best? Sorry about the our lack of knowledge, we Thunderbird developers have to interface with many parts of Mozilla Core we have little or no knowledge of. Would there be a security benefit of porting the Firefox changes instead of disabling the test? I see that in services/blocklists/moz.build you added +if CONFIG['MOZ_BUILD_APP'] == 'browser': + DIST_SUBDIR = 'browser' which may relate to +@RESPATH@/browser/defaults/blocklists Sadly my forte is also not build configuration, but we could get some of our build guys to look into porting this.
> Can you please advise us on what is best? I would suggest to skip the test to get your green build back. And maybe open a new issue to mimic the delivery of those JSON files. > Sorry about the our lack of knowledge Please, don't be sorry! Basically, the Bug 1224528 consists in shipping initial JSON files for the blocklist settings, so that clients do not have to download the whole list on a cold start (first use). > Would there be a security benefit of porting the Firefox changes instead of disabling the test? AFAIK, no.
Thanks. We've already been trough bug 1266794, bug 1247662, bug 1248557 and bug 1312827 in this area. Hopefully one day we'll have someone knowable enough on the TB team to decide TB's way forward on this.
Pushed by email@example.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/185d93157773 Skip test_blocklist_certificates.js for Thunderbird. r=leplatrem