Closed Bug 1380605 Opened 5 years ago Closed 4 years ago

Provide OpenH264 v1.7.1 build


(Release Engineering :: Release Requests, enhancement, P1)



(Not tracked)



(Reporter: nthomas, Assigned: nthomas)




(7 files)

Time for another OpenH264 release, last time was bug 1286533.
Notes from investigating
* bug 1284285 comment #1 the download domain is going to change from to 
 --> TODO we'll need to adjust the Balrog whitelist

* we last did this about a year ago, bug 1286533 comment #6 has the steps Callek used
* 52.0 was the last release cycle that support 32-bit mac, only need 64-bit if this is going to ride the trains with 55/56
* we have an AArch64 for android now
* our dump_syms executables for mac/linux64/windows are from Dec 2015, and there's been a couple of syncs with upstream since then, so we should refresh them (possibly by just grabbing from an appropriate build and put them in tooltool as public artifacts)
* expecting to have to iterate to find which parts of main tooltool manifests are necessary these days, eg  tooltool manifests for desktop don't include gcc/clang/visual studio/rustc
* cross-compiling on mac might be fun, for arbitrarily small values of fun
- first updates of openh264 script and config at try 0619db083bef (based on attachment 8770575 [details])
- using that & openh264 rev 6c2e7008f8bbe2ff90100972f97071eb87ec37cb (tip of v1.7.1-Firefox39)
- linux32 and linux64 compile & test without error
- android builds fail with an ndk issue (I've updated to the r11c ndk to match desktop), will debug this next
Attachment #8892697 - Flags: review?(bhearsum)
Commit pushed to master at
Bug 1380605 - add new OpenH264 download host to whitelists (#363). r=bhearsum
(In reply to Nick Thomas [:nthomas] from comment #3)
> Created attachment 8892697 [details] [review]
> PR to add new download host

Do you know what the timeline for this release is? We normally avoid doing code pushes during merge & release weeks. If we want or need to ship prior to August 16th please let me know, and I'll ensure we get a push done.
Flags: needinfo?(nthomas)
Attachment #8892697 - Flags: review?(bhearsum) → review+
We won't need a push before then.
Flags: needinfo?(nthomas)
Hi Nick,
Our PE Mo ( just pointed out the new domain name ( to host the binaries is the internal container name of the CDN, which should not be used by the users directly. There would be problem if the CDN provider changes in the future. 

We are trying to fix it right now. Please hold on your release until we fix this issue. Thanks.
Flags: needinfo?(nthomas)
Flags: needinfo?(mzanaty)
OK, waiting for more information.
Flags: needinfo?(nthomas)
Attachment #8893584 - Flags: review?(bhearsum)
Commit pushed to master at
Revert "Bug 1380605 - add new OpenH264 download host to whitelists (#363). r=bhearsum" (#365)
Attachment #8893584 - Flags: review?(bhearsum) → review+
See Also: → 1225488
Hi Nick,
We managed to keep the previous CDN domain name ( and finished the deployment. You can go ahead with your release and send the plugin binaries to me when ready. Thanks.
Flags: needinfo?(mzanaty)
Depends on: 1388901
Depends on: 1390024
Status update - the builds are nearly ready, I'll sign them tomorrow and pass them over to Cisco.
Attached file Docs
Builds are generated and emailed to Hank at Cisco

Docs attached are here. Our code reference is try a8125131fc56cb50ff1ede4e800dd3dcdac0c8f3, tracing back to mozilla-central 933a04a91ce3bd44b230937083a835cb60637084. I'll put the patch up for review so we can get it into the tree.
Attached file Zip of log files
Hi Nick,
All the files are uploaded into CDN now. Here are the download links, please check out.
Flags: needinfo?(nthomas)
Thanks. I've set up our update server on the nightlytest channel to serve 1.7.1. I'm not 100% sure sure I have the build target correct for aarch64 (bug 1393312).

Note to self - once we ship to nightly we should update
Flags: needinfo?(nthomas)
Attached file
The json for the balrog release was created by hand (could do with a script for that), so this is a quick hack to verify that all the urls work, as well as the file size and hash matching.
Comment on attachment 8902942 [details]
Bug 1380605 - modifications to build openh264 v1.7.1

Blurk, my plans to have one review request failed miserably. Just let me know if you'd rather have one roll-up patch instead.
Attachment #8902942 - Flags: review?(catlee)
Comment on attachment 8902940 [details]
Bug 1380605 - Rebased patch from OpenH264 v1.6 (bug 1286533), p=Callek
Attachment #8902940 - Flags: review?(catlee) → review+
Comment on attachment 8902942 [details]
Bug 1380605 - modifications to build openh264 v1.7.1
Attachment #8902942 - Flags: review?(catlee) → review+
Pushed by
Rebased patch from OpenH264 v1.6 (bug 1286533), r=catlee
modifications to build openh264 v1.7.1, r=catlee
Keywords: leave-open
When this bug landed, these improvements have been noticed for about 3 days.
I'm commenting here for easier perf tracking.

== Change summary for alert #9342 (as of September 08 2017 04:53 UTC) ==


 11%  remote-blank summary android-4-4-armv7-api16 opt      1,110.85 -> 993.52

For up to date results, see:
(In reply to Ionuț Goldan [:igoldan], Performance Sheriffing from comment #25)

I think it must have been something else, the code here doesn't affect Fennec builds at all.
Tracking deployment in bug 1284285.
Closed: 4 years ago
Resolution: --- → FIXED
Removing leave-open keyword from resolved bugs, per :sylvestre.
Keywords: leave-open
Component: Custom Release Requests → Release Requests
You need to log in before you can comment on or make changes to this bug.