As part of the Dawn project, we're moving DevEdition builds from mozilla-aurora to mozilla-beta. This means that we'll now have two shipping build configurations built from the same branch. We need to continue using a separate MAR_CHANNEL_ID for DevEdition (both to avoid accidents and make it impossible for a MitM to move a Beta user to DevEdition build, which has unsigned addons enabled). I spoke with nalexander about this a bit and he had some suggestions on how to go about it: 16:10 < nalexander> bhearsum: did you get an answer? In short, you can do this in the Python configure, I expect; I'd talk to chmanchester about that. 16:11 < bhearsum> nalexander: i did not 16:11 < nalexander> bhearsum: let me see if I can find an example. 16:11 < bhearsum> that would be awesome 16:12 < nalexander> bhearsum: https://dxr.mozilla.org/mozilla-central/source/mobile/android/moz.configure#15 suggests the approach. 16:12 < nalexander> bhearsum: now, it's possible the MAR channel IDs are special, let me try to evalute that. 16:14 < nalexander> bhearsum: I don't see anything that screams "this needs to be old-configure.in" and not in moz.configure Python; I expect you can use something like that stanza (expanding the is_nightly/is_... as necessary) to achieve what you want. 16:16 < bhearsum> nalexander: does that mean i should remove them from confvars.sh and put them into https://dxr.mozilla.org/mozilla-beta/source/browser/moz.configure ? 16:17 < nalexander> bhearsum: yes, just so -- remove the old-configure.in bits, remove the confvars.sh setting, add to browser/moz.configure (since it seems to be browser/ specific). 16:18 < nalexander> bhearsum: for testing, you can do a fresh configure with `rm -f $OBJDIR/config.status && ./mach configure`; then you can examine `$OBJDIR/config.cache` to see the values produced. 16:19 < nalexander> bhearsum: ./mach environment -v | grep CHANNEL_ID does roughly the config.cache checking too.

I originally started down the road of moving this to moz.configure but I bailed after hitting some issues, and realizing that it probably also has implications for all gecko apps. This is a simplest solution to the immediate problem.

Comment on attachment 8864140 [details] [diff] [review] set special MAR_CHANNEL_ID and ACCEPTED_MAR_CHANNEL_IDS for DevEdition Review of attachment 8864140 [details] [diff] [review]: ----------------------------------------------------------------- I expect this to be fine, but redirecting to chmanchester for final stamp.

Comment on attachment 8864140 [details] [diff] [review] set special MAR_CHANNEL_ID and ACCEPTED_MAR_CHANNEL_IDS for DevEdition Review of attachment 8864140 [details] [diff] [review]: ----------------------------------------------------------------- This code looks reasonable but I'm sorry to say I don't know enough about the updater to know whether this is the right way to achieve this.

Comment on attachment 8864140 [details] [diff] [review] set special MAR_CHANNEL_ID and ACCEPTED_MAR_CHANNEL_IDS for DevEdition Review of attachment 8864140 [details] [diff] [review]: ----------------------------------------------------------------- bhearsum: based on the patch and my interpretation of http://searchfox.org/mozilla-central/source/testing/mozharness/configs/merge_day/central_to_aurora.py#49, feel free to put my r+ on it, but I'm not expert in this area. I'm asking for review from rstrong, who I expect knows a great deal about such things and can redirect yet further if needed.

(In reply to Nick Alexander :nalexander from comment #4) > Comment on attachment 8864140 [details] [diff] [review] > set special MAR_CHANNEL_ID and ACCEPTED_MAR_CHANNEL_IDS for DevEdition > > Review of attachment 8864140 [details] [diff] [review]: > ----------------------------------------------------------------- > > bhearsum: based on the patch and my interpretation of > http://searchfox.org/mozilla-central/source/testing/mozharness/configs/ > merge_day/central_to_aurora.py#49, feel free to put my r+ on it, but I'm not > expert in this area. FYI, this config (and the aurora_to_beta.py) will get replaced by a new central_to_beta.py one (bug 1360542) which will change the values in the else block to the correct values for Beta. I expect the aurora side of this to continue to exist on Beta and Release, but go unused due to nothing setting it as the channel. > I'm asking for review from rstrong, who I expect knows > a great deal about such things and can redirect yet further if needed. Sounds good. Robert, let me know if you need any additional context on this.

Comment on attachment 8864140 [details] [diff] [review] set special MAR_CHANNEL_ID and ACCEPTED_MAR_CHANNEL_IDS for DevEdition This should be fine with corresponding changes to the merge day scripts.

Thanks for all of the reviews, folks. I'll be landing this for additional testing on jamun before it heads to inbound and beta.

Pushed by bhearsum@mozilla.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/60175653dc7b allow MAR_CHANNEL_ID and ACCEPTED_MAR_CHANNEL_IDS to be set different for DevEdition. r=nalexander,rstrong

Comment on attachment 8864140 [details] [diff] [review] set special MAR_CHANNEL_ID and ACCEPTED_MAR_CHANNEL_IDS for DevEdition Landed on inbound and beta (we've got carte blanche to land Dawn-related things without prior approval): remote: https://hg.mozilla.org/integration/mozilla-inbound/rev/60175653dc7b3966cf7df0e4a02dc7305051e839 remote: https://hg.mozilla.org/releases/mozilla-beta/rev/93700f9bdc9cb4e278d331049dff765c75771632