Opened 6 years ago

Closed 6 years ago

Last modified 6 years ago

#10159 closed defect (fixed)

omni.ja differs in ESR24 Gitian builds

Reported by: gk Owned by: erinn
Priority: Medium Milestone:
Component: Applications/Tor bundles/installation Version:
Severity: Keywords: ff24-esr, tbb-3.0, tbb-testcase, MikePerry201311R
Cc: mikeperry Actual Points:
Parent ID: #10103 Points:
Reviewer: Sponsor:

Description

Testing two builds with the new OS X cross-compiler showed that browser/omni.ja differed. More exactly, only the contents of chrome.manifest showed some reordering.

Child Tickets

Attachments (2)

omnidiff (1.0 KB) - added by gk 6 years ago.
0001-bug-10159-making-the-manifest-files-deterministic-by.patch (759 bytes) - added by gk 6 years ago.

Download all attachments as: .zip

Change History (14)

comment:1 Changed 6 years ago by gk

Parent ID: #10103

comment:2 Changed 6 years ago by gk

Owner: changed from #10103 to erinn
Status: newassigned

comment:3 Changed 6 years ago by gk

Status: assignednew

Changed 6 years ago by gk

Attachment: omnidiff added

comment:4 Changed 6 years ago by gk

I investigated the omni.ja packaging step and nothing jumped out at me that could explain the diff in the omnidiff attachment. Thus, I run 20 package tests on Linux and 20 on Mac and they produced always the same omni.ja. My new idea is that the step generating the browser.manifest which in turn gets merged into the problematic chrome.manifest during omni.ja packaging is the culprit here. Looking into that now.

comment:5 Changed 6 years ago by gk

Yes. I ran two builds on my test Mac and they produced different browser.manifest files.

comment:6 Changed 6 years ago by gk

Hmm, running JarMaker.py two hundred times did not help in reproducing the problem. Seems there is still more to it.

Last edited 6 years ago by gk (previous) (diff)

comment:7 Changed 6 years ago by gk

Okay. Just rebuilding the browser part does already trigger the problem. This takes much less than a minute and I can start debugging this seriously. The other option that is helping us is building with -j1 (according to Mike Hommey) which I'd avoid if I could.

comment:8 Changed 6 years ago by gk

Status: newneeds_review

Finally, after a lot of pain...

comment:9 Changed 6 years ago by mikeperry

Keywords: MikePerry201311R added

comment:10 Changed 6 years ago by mikeperry

Resolution: fixed
Status: needs_reviewclosed

Ok. Merged as a fixup commit in tor-browser-24.1.1esr-1.

Do we also want the javascript patch from https://bugzilla.mozilla.org/show_bug.cgi?id=943331#c4, too? Reopen and make sure this is re-tagged to be on my radar if so.

comment:11 in reply to:  10 Changed 6 years ago by gk

Replying to mikeperry:

Ok. Merged as a fixup commit in tor-browser-24.1.1esr-1.

Do we also want the javascript patch from https://bugzilla.mozilla.org/show_bug.cgi?id=943331#c4, too?

Yes, in #10235. I'll attach a patch shortly.

Last edited 6 years ago by gk (previous) (diff)

comment:12 Changed 6 years ago by gk

Keywords: tbb-testcase added
Note: See TracTickets for help on using tickets.