Open Bug 1502457 Opened 6 years ago Updated 2 years ago

Contextual Feature Recommender (CFR) and other "nightly/local"-only features uses "default" to select cohort

Categories

(Firefox Build System :: General, defect)

defect

Tracking

(Not tracked)

REOPENED
mozilla65

People

(Reporter: mozilla, Assigned: glandium)

References

Details

Attachments

(6 files)

I briefly saw the blue "recommendation" button for (what I believe to be) the Contextual Feature Recommender (or its extension-recommending subcomponent??) in the URL bar on Firefox 63 today, despite having studies disabled (and about:studies being empty).

It appeared right before I clicked away from the page in question, and didn't return when I went back to the same page.

After that, there was no sign of it that I could find  (e.g. in the page info dialogue).

I eventually tracked it down to being CFR only after image-searching for likely phrases like "firefox recommendation blue url button".

The only reference I can find to this feature is in about:config in browser.newtabpage.activity-stream.asrouter.messageProviders, which includes {"id":"cfr","type":"local","localProvider":"CFRMessageProvider","enabled":true,"cohort":"nightly"}]. I am not on nightly.

So this seems to be a feature to recommend extensions for certain sites?

There are some serious issues with this experience:

* I see a new browser feature appear momentarily. The fact that it is shown on a specific site suggests that Firefox *might* be sending info about my browsing history to Mozilla, depending on the feature's implementation. But then it disappears completely, making it a PITA to find out what it was. The feature is not documented at https://www.mozilla.org/en-US/firefox/63.0/releasenotes/ either.
* If this isn't a general-release feature, I'm specifically opted out of Shield studies, so I shouldn't be seeing it either.
* There seems to be FA documentation about how this works. Local/generic domain->recommended extension map? All visited domains go to some Mozilla Shield study server somewhere? The only info I could find really is https://techdows.com/2018/08/firefox-to-show-extension-recommendations.html . Why is this seemingly not documented anywhere by Mozilla? This makes things look *worse*.
* How do I switch this off?? The above page (of unknown reliability) mentions toggling a pref that doesn't exist by default.

To address this, the following should happen at a *minimum*:

* The button re-appears when the user returns to the domain or site (or shows up on the URL bar LHS as an icon).
* Clicking that button/icon shows at least: "switch this feature off completely" and "how does this work/privacy info".
* The feature should be *documented* by Mozilla somewhere.

There is also bigger meta-issues here: Mozilla as a development organisation keeps making the same mistakes again and again in relation to privacy-connected changes (see: Mr. Robot fiasco). There should be a set of standard minimum requirements/checks when a new feature has or *appears to have* a privacy implication. If this exists, it's clearly not working.
I discovered https://support.mozilla.org/en-US/kb/extension-recommendations after trawling r/firefox.

There are two problems with that page:

"Mozilla does not collect or store your browsing history to make those recommendations."

Does anyone else? What constitutes "browsing history"? Is *anything* sent (domains?), or is this purely local (comparing locally to a generic list)?

"How do I opt out of recommendations in Firefox?"

This is clearly an obvious or common enough question to be included on that page, but there is *no option* to actually opt out. Dismissing individual recommendations is *not* the same as opting out of recommendations.
Status: UNCONFIRMED → NEW
Ever confirmed: true
Thanks for the report :moz-nuc.  

This feature was turned on for 30% of the beta 63 population and is pref'd off by default in 63 release.  Which channel where/are you on when you saw the recommendation?  

If you had Shield studies turned off it should have been impossible to see a recommendation, and this would be a bug.  Could you go into about:config and search for the pref `browser.newtabpage.activity-stream.asrouter.messageProviders` and paste the value in a comment?  

This feature does not affect users' privacy and shares no data with third party sites.  Firefox is recommending an extension to the user based how often the user visits the corresponding site, and will trigger the recommendation when the user is currently visiting the site.  The site remains unaware of the recommendation and no data is exchanged.  The idea of this feature is to recommend extensions to user that have proven to improve the user's experience 

The feature hasn't been released yet, so isn't documented in a user facing way.  It is an experiment, so is documented in bugzilla and internal documents.  Here's the metabug:  bug #1471328 and this bug is the Shield experiment: bug #1491379

The 'switch off' feature has landed in Nightly 65 and will allow you to opt out of the feature or opt out of specific recommendations.
Flags: needinfo?(mozilla)
Thanks for the quick/detailed response; appreciated :)

> Which channel where/are you on

Release, via Fedora - I've never used beta/nightly channels. Fedora SRPM (i.e. upstream source + Fedora build config.) is here: https://kojipkgs.fedoraproject.org//packages/firefox/63.0/2.fc28/src/firefox-63.0-2.fc28.src.rpm

> browser.newtabpage.activity-stream.asrouter.messageProviders

See my original comment - CFR is included in the above key. Full value:

[{"id":"onboarding","type":"local","localProvider":"OnboardingMessageProvider","enabled":false,"cohort":0},{"id":"snippets","type":"remote","url":"https://snippets.cdn.mozilla.net/us-west/bundles/bundle_d6d90fb9098ce8b45e60acf601bcb91b68322309.json","updateCycleInMs":14400000,"enabled":false},{"id":"cfr","type":"local","localProvider":"CFRMessageProvider","enabled":true,"cohort":"nightly"}]

> This feature...

Thanks for the further info. I understand, having dug into it more, that the functionality is client-side based on a locally-stored domain/extension map, but it's not clear from the current UX.

> it should have been impossible to see a recommendation

Unfortunately, it subsequently happened on a second machine/profile too, so definitely not an isolated case.
Flags: needinfo?(mozilla)
Thanks, :moz-nuc.

One more question about the browser.newtabpage.activity-stream.asrouter.messageProviders pref.  Could you take a look at the "Status" column of that pref?  If CFR has "enabled":true, then the status would have to "modified".  If it's "default", then that would be pretty weird.  In the case that it is "default", I would ask that you create a new profile and then take a look at the pref again and let me know what the values of the Status and Value columns are.
Flags: needinfo?(mozilla)
That's the default value.

And from a new profile:

[{"id":"onboarding","type":"local","localProvider":"OnboardingMessageProvider","enabled":false,"cohort":0},{"id":"snippets","type":"remote","url":"https://snippets.cdn.mozilla.net/us-west/bundles/bundle_d6d90fb9098ce8b45e60acf601bcb91b68322309.json","updateCycleInMs":14400000,"enabled":false},{"id":"cfr","type":"local","localProvider":"CFRMessageProvider","enabled":true,"cohort":"nightly"}]
Flags: needinfo?(mozilla)
If you are on firefox 63 and are seeing CFR enabled as a default value it might be because you have an unofficial build. Did you build your own firefox locally or downloaded it from the website?
Can you please post your Build ID from the `about:support` page?
Also if you could open your browser toolbox [0] and share the value of `AppConstants.MOZILLA_OFFICIAL` it would help us figure out this issue.
Thanks.


[0] https://developer.mozilla.org/en-US/docs/Tools/Browser_Toolbox
Flags: needinfo?(mozilla)
> Did you build your own firefox locally or downloaded it from the website?

Answered above:

> Release, via Fedora - I've never used beta/nightly channels. Fedora SRPM (i.e. upstream source + Fedora build config.) is here: https://kojipkgs.fedoraproject.org//packages/firefox/63.0/2.fc28/src/firefox-63.0-2.fc28.src.rpm

> AppConstants.MOZILLA_OFFICIAL

true
Flags: needinfo?(mozilla)
First off: moz-nuc, thank you for your patience and tenacity here, bug reports like this are a huge help.
Summary: Contextual Feature Recommender (CFR) terrible user experience → Contextual Feature Recommender (CFR) uses "default" to select cohort.
Second: From a casual investigation, I suspect the issue here is is that we have overloaded the term "default", and unintentionally selected from our release population on partner distributions as a result. 

In: chrome/browser/res/activity-stream/lib/ActivityStream.jsm

const IS_NIGHTLY_OR_UNBRANDED_BUILD = ["nightly", "default"].includes(AppConstants.MOZ_UPDATE_CHANNEL);

I suspect we're accidentally pulling in members of the wrong cohort - including yourself - as a result of this.

I'm going to investigate further, but I've asked the owner of this RPM over at RedHat to join us in this bug, and if this turns out to be the issue, to ask him if he can respin those builds.
This has been a long-standing problem with no good solution :(  I had to deal with something similar in the e10s and flash click-to-play rollouts.

You can change from checking AppConstants.MOZ_UPDATE_CHANNEL to directly checking AppConstants.NIGHTLY_BUILD, which will give the expected results:
 - true for nightly or local builds
 - false for release builds of Fedora, for example.


If for some reason you must remain using the MOZ_UPDATE_CHANNEL, take a look here: https://hg.mozilla.org/releases/mozilla-beta/file/94cf7f1ee06b/browser/extensions/clicktoplay-rollout/bootstrap.js#l22 (noting that this is solving the problem on the opposite direction)
Sure, I'm here to help. As the update, Fedora builds the packages with those (related) options:

export BUILD_OFFICIAL=1
export MOZILLA_OFFICIAL=1
export MOZ_TELEMETRY_REPORTING=1
mk_add_options BUILD_OFFICIAL=1
mk_add_options MOZILLA_OFFICIAL=1
ac_add_options --enable-official-branding
ac_add_options --disable-updater
Story recommendation v2 is also using the same IS_NIGHTLY_OR_UNBRANDED_BUILD value although that code isn't in 63 but will be in 64 "nightly" (bug 1498478)
Component: Shield Study → Activity Streams: Newtab
Product: Shield → Firefox
See Also: → 1498478
Summary: Contextual Feature Recommender (CFR) uses "default" to select cohort. → Contextual Feature Recommender (CFR) and other "nightly/local"-only features uses "default" to select cohort
(In reply to Martin Stránský [:stransky] from comment #11)
> Sure, I'm here to help. As the update, Fedora builds the packages with those
> (related) options:
> 
> export BUILD_OFFICIAL=1
> export MOZILLA_OFFICIAL=1
> export MOZ_TELEMETRY_REPORTING=1
> mk_add_options BUILD_OFFICIAL=1
> mk_add_options MOZILLA_OFFICIAL=1
> ac_add_options --enable-official-branding
> ac_add_options --disable-updater

Would an `ac_add_options --update-channel release` help here for correctly identifying these fedora builds as release?
(In reply to William Lachance (:wlach) (use needinfo!) from comment #13)
> Would an `ac_add_options --update-channel release` help here for correctly
> identifying these fedora builds as release?

(note that I myself am not 100% sure of the implications of changing this setting, so please don't do this without checking with someone more knowledgable)
Component: Activity Streams: Newtab → General
(In reply to William Lachance (:wlach) (use needinfo!) from comment #14)
> (In reply to William Lachance (:wlach) (use needinfo!) from comment #13)
> > Would an `ac_add_options --update-channel release` help here for correctly
> > identifying these fedora builds as release?
> 
> (note that I myself am not 100% sure of the implications of changing this
> setting, so please don't do this without checking with someone more
> knowledgable)

ac_add_options --update-channel=release seems to do the trick, Thanks! The variable is now:

[{"id":"onboarding","type":"local","localProvider":"OnboardingMessageProvider","enabled":false,"cohort":0},{"id":"snippets","type":"remote","url":"https://snippets.cdn.mozilla.net/us-west/bundles/bundle_d6d90fb9098ce8b45e60acf601bcb91b68322309.json","updateCycleInMs":14400000,"enabled":false},{"id":"cfr","type":"local","localProvider":"CFRMessageProvider","enabled":false,"cohort":""}]
(In reply to Mike Hoye [:mhoye] from comment #9)
> Second: From a casual investigation, I suspect the issue here is is that we
> have overloaded the term "default", and unintentionally selected from our
> release population on partner distributions as a result. 
> 
> In: chrome/browser/res/activity-stream/lib/ActivityStream.jsm
> 
> const IS_NIGHTLY_OR_UNBRANDED_BUILD = ["nightly",
> "default"].includes(AppConstants.MOZ_UPDATE_CHANNEL);

This is wrong on so many levels :( /That/ is what should be fixed, including in mozilla-central.
Or... considering how many feet have been shot already, automatically set MOZ_UPDATE_CHANNEL depending on the milestone and whether developer options are set.
(In reply to Mike Hommey [:glandium] from comment #17)
> Or... considering how many feet have been shot already, automatically set
> MOZ_UPDATE_CHANNEL depending on the milestone and whether developer options
> are set.

--> Fx build system. :glandium, can you take this?
Flags: needinfo?(mh+mozilla)
Product: Firefox → Firefox Build System
BTW downstream has now added a changelog to disable all(?) Firefox studies… I think this is not reasonable as – if it does, what it says – just totally prevents any studies that may influence Firefox development on a whole distro. Also the about:studies site is broken. Opened https://bugzilla.redhat.com/show_bug.cgi?id=1648632 for both.
(In reply to rugk from comment #19)
> BTW downstream has now added a changelog to disable all(?) Firefox studies…
> I think this is not reasonable as – if it does, what it says – just totally
> prevents any studies that may influence Firefox development on a whole
> distro. Also the about:studies site is broken. Opened
> https://bugzilla.redhat.com/show_bug.cgi?id=1648632 for both.

As an ordinary(-ish) user on Fedora, I have to say that Mozilla has made a *lot* of questionable decisions about user privacy and crapware shovelling lately (Cliqz? Mr. Robot advertising?), and I'm pretty sympathetic at the moment if Fedora has decided to disable studies entirely, until and unless Mozilla gets its house in order.
Indeed they did not do everything right in the past, but they have apologized for that Mr. Robot thing (https://www.cnet.com/news/mozilla-investigates-mr-robot-firefox-extension-problem/) and promised not to do it again.

Anyway, whether to enable or disable that thing must stay in the user's hand. Also this is getting off-topic, as this issue is not about studies in general, but about a specific bug in the user selection that affected Fedora users.
Assignee: nobody → mh+mozilla
Flags: needinfo?(mh+mozilla)
Pushed by mh@glandium.org:
https://hg.mozilla.org/integration/autoland/rev/638fa0251123
Use MOZ_APP_VERSION instead of FIREFOX_VERSION. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/0e283ddc6c3b
Move FIREFOX_VERSION to python configure. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/b172bda6e983
Move MOZ_APP_VERSION{,_DISPLAY} to python configure. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/73bb6a74cad0
Move js_option around to make it available earlier. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/7ea52c602445
Move MOZILLA_OFFICIAL to init.configure. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/febdb867785a
Derive the default update channel from the application display version. r=nalexander
Ugh, those windows failures are a demonstration of what's wrong with the current use of MOZ_UPDATE_CHANNEL: per-push builds are passed --enable-update-channel=, which currently means "default". Let's keep it that way for now, but I'll file a followup bug.
Flags: needinfo?(mh+mozilla)
Pushed by mh@glandium.org:
https://hg.mozilla.org/integration/autoland/rev/2c2b09264a82
Use MOZ_APP_VERSION instead of FIREFOX_VERSION. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/a73cfbabdd47
Move FIREFOX_VERSION to python configure. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/1cf7694c8aa2
Move MOZ_APP_VERSION{,_DISPLAY} to python configure. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/1442604ea266
Move js_option around to make it available earlier. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/2bc4c0bfa299
Move MOZILLA_OFFICIAL to init.configure. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/9174dadf38b0
Derive the default update channel from the application display version. r=nalexander
Backed out 6 changesets (Bug 1502457) for causing Mn mozprocess automation failures in '/Users/cltbld/tasks/task_1542517470/build/venv/bin/python

https://treeherder.mozilla.org/#/jobs?repo=autoland&fromchange=eed15f1e00944954d89865708a7bf7e759fe3d50&group_state=expanded&searchStr=os%2Cx%2C10.10%2Cdebug%2Ctest-macosx64%2Fdebug-marionette-e10s%2C%28mn%29&tochange=9cfcc27a9446824df7777bd93b2a4878e348098d&selectedJob=212472775
https://treeherder.mozilla.org/logviewer.html#?job_id=212472775&repo=autoland&lineNumber=21959

22:04:57     INFO - TEST-START | testing/marionette/harness/marionette_harness/tests/unit/test_timeouts.py TestTimeouts.test_reset_timeout
22:04:57     INFO -  1542521097389	Marionette	TRACE	53 <- [1,10,null,{"value":null}]
22:04:57     INFO -  1542521097390	Marionette	DEBUG	Closed connection 53
22:04:57     INFO -  1542521097391	Marionette	DEBUG	Accepted connection 54 from 127.0.0.1:51809
22:04:57     INFO -  1542521097392	Marionette	TRACE	54 -> [0,1,"WebDriver:NewSession",{"strictFileInteractability":true}]
22:04:57     INFO -  1542521097394	Marionette	DEBUG	[2147483778] Frame script loaded
22:04:57     INFO -  1542521097394	Marionette	DEBUG	[2147483778] Frame script registered
22:04:57     INFO -  1542521097395	Marionette	TRACE	54 <- [1,1,null,{"sessionId":"56ab5e74-0868-f540-90c3-e13fb641f443","capabilities":{"browserName":"firefox","browserVersion":"65.0a ... /T/tmpsYxSAH.mozrunner","moz:shutdownTimeout":60000,"moz:useNonSpecCompliantPointerOrigin":false,"moz:webdriverClick":true}}]
22:04:57     INFO -  1542521097396	Marionette	TRACE	54 -> [0,2,"WebDriver:SetTimeouts",{"script":30000}]
22:04:57     INFO -  1542521097397	Marionette	TRACE	54 <- [1,2,null,{"value":null}]
22:04:57     INFO -  1542521097398	Marionette	TRACE	54 -> [0,3,"WebDriver:SetTimeouts",{"pageLoad":300000}]
22:04:57     INFO -  1542521097399	Marionette	TRACE	54 <- [1,3,null,{"value":null}]
22:04:57     INFO -  1542521097402	Marionette	TRACE	54 -> [0,4,"WebDriver:SetTimeouts",{"implicit":0}]
22:04:57     INFO -  1542521097402	Marionette	TRACE	54 <- [1,4,null,{"value":null}]
22:04:57     INFO -  1542521097403	Marionette	TRACE	54 -> [0,5,"WebDriver:GetTimeouts",{}]
22:04:57     INFO -  1542521097403	Marionette	TRACE	54 <- [1,5,null,{"implicit":0,"pageLoad":300000,"script":30000}]
22:04:57     INFO -  1542521097404	Marionette	TRACE	54 -> [0,6,"WebDriver:GetTimeouts",{}]
22:04:57     INFO -  1542521097404	Marionette	TRACE	54 <- [1,6,null,{"implicit":0,"pageLoad":300000,"script":30000}]
22:04:57     INFO -  1542521097405	Marionette	TRACE	54 -> [0,7,"WebDriver:GetTimeouts",{}]
22:04:57     INFO -  1542521097405	Marionette	TRACE	54 <- [1,7,null,{"implicit":0,"pageLoad":300000,"script":30000}]
22:04:57     INFO -  1542521097406	Marionette	TRACE	54 -> [0,8,"Marionette:GetContext",{}]
22:04:57     INFO -  1542521097406	Marionette	TRACE	54 <- [1,8,null,{"value":"content"}]
22:04:57     INFO -  1542521097407	Marionette	TRACE	54 -> [0,9,"WebDriver:DeleteSession",{}]
22:04:57     INFO -  1542521097408	Marionette	TRACE	54 <- [1,9,null,{"value":null}]
22:04:57     INFO -  1542521097409	Marionette	DEBUG	Closed connection 54
22:21:37     INFO - Automation Error: mozprocess timed out after 1000 seconds running ['/Users/cltbld/tasks/task_1542517470/build/venv/bin/python', '-u', '/Users/cltbld/tasks/task_1542517470/build/tests/marionette/harness/marionette_harness/runtests.py', '--gecko-log=-', '--log-raw=-', '-vv', '--log-raw=/Users/cltbld/tasks/task_1542517470/build/blobber_upload_dir/marionette_raw.log', '--log-errorsummary=/Users/cltbld/tasks/task_1542517470/build/blobber_upload_dir/marionette_errorsummary.log', '--log-html=/Users/cltbld/tasks/task_1542517470/build/blobber_upload_dir/report.html', '--binary=/Users/cltbld/tasks/task_1542517470/build/application/Firefox NightlyDebug.app/Contents/MacOS/firefox', '--address=localhost:2828', '--symbols-path=/Users/cltbld/tasks/task_1542517470/build/symbols', '/Users/cltbld/tasks/task_1542517470/build/tests/marionette/tests/testing/marionette/harness/marionette_harness/tests/unit-tests.ini']
22:21:37    ERROR - timed out after 1000 seconds of no output
22:21:37    ERROR - Return code: -15
22:21:37    ERROR - No suite end message was emitted by this harness.
22:21:37     INFO - TinderboxPrint: marionette<br/>409/0/1
22:21:37     INFO - gecko.log not found
22:21:37     INFO - TinderboxPrint: marionette<br/>409/0/1
22:21:37     INFO - Marionette exited with return code -15: FAILURE
22:21:37    ERROR - # TBPL FAILURE #
22:21:37  WARNING - setting return code to 2
22:21:37     INFO - Running post-action listener: _package_coverage_data
22:21:37     INFO - Running post-action listener: _resource_record_post_action
22:21:37     INFO - Running post-action listener: process_java_coverage_data
22:21:37     INFO - [mozharness: 2018-11-18 06:21:37.453981Z] Finished run-tests step (success)
22:21:37     INFO - Running post-run listener: _resource_record_post_run
22:21:37     INFO - Total resource usage - Wall time: 1293s; CPU: 41.0%; Read bytes: 44630016; Write bytes: 893559296; Read time: 334; Write time: 2817
22:21:37     INFO - TinderboxPrint: CPU usage<br/>41.1%
22:21:37     INFO - TinderboxPrint: I/O read bytes / time<br/>44,630,016 / 334
22:21:37     INFO - TinderboxPrint: I/O write bytes / time<br/>893,559,296 / 2,817
22:21:37     INFO - TinderboxPrint: CPU idle<br/>3,046.0 (58.9%)
22:21:37     INFO - TinderboxPrint: CPU system<br/>1,176.6 (22.8%)
22:21:37     INFO - TinderboxPrint: CPU user<br/>947.5 (18.3%)
22:21:37     INFO - TinderboxPrint: Swap in / out<br/>306,851,840 / 0
22:21:37     INFO - install - Wall time: 24s; CPU: 54.0%; Read bytes: 203048960; Write bytes: 205231616; Read time: 18264; Write time: 407
22:21:37     INFO - run-tests - Wall time: 1269s; CPU: 41.0%; Read bytes: 43753472; Write bytes: 669797376; Read time: 322; Write time: 2369
22:21:38  WARNING - returning nonzero exit status 2
[taskcluster 2018-11-18T06:21:38.130Z]    Exit Code: 2
[taskcluster 2018-11-18T06:21:38.130Z]    User Time: 4m15.445968s
[taskcluster 2018-11-18T06:21:38.130Z]  Kernel Time: 44.724993s
[taskcluster 2018-11-18T06:21:38.130Z]    Wall Time: 23m4.816780658s
[taskcluster 2018-11-18T06:21:38.130Z]       Result: FAILED
[taskcluster 2018-11-18T06:21:38.130Z] === Task Finished ===
[taskcluster 2018-11-18T06:21:38.130Z] Task Duration: 23m4.816977616s
[taskcluster 2018-11-18T06:21:38.921Z] Uploading artifact public/logs/localconfig.json from file logs/localconfig.json with content encoding "gzip", mime type "application/json" and expiry 2019-11-18T05:58:28.630Z
[taskcluster 2018-11-18T06:21:39.471Z] Uploading artifact public/logs/log_critical.log from file logs/log_critical.log with content encoding "gzip", mime type "text/plain" and expiry 2019-11-18T05:58:28.630Z
[taskcluster 2018-11-18T06:21:39.827Z] Uploading artifact public/logs/log_error.log from file logs/log_error.log with content encoding "gzip", mime type "text/plain" and expiry 2019-11-18T05:58:28.630Z
[taskcluster 2018-11-18T06:21:40.252Z] Uploading artifact public/logs/log_fatal.log from file logs/log_fatal.log with content encoding "gzip", mime type "text/plain" and expiry 2019-11-18T05:58:28.630Z
[taskcluster 2018-11-18T06:21:40.617Z] Uploading artifact public/logs/log_info.log from file logs/log_info.log with content encoding "gzip", mime type "text/plain" and expiry 2019-11-18T05:58:28.630Z
[taskcluster 2018-11-18T06:21:41.315Z] Uploading artifact public/logs/log_raw.log from file logs/log_raw.log with content encoding "gzip", mime type "text/plain" and expiry 2019-11-18T05:58:28.630Z
[taskcluster 2018-11-18T06:21:41.961Z] Uploading artifact public/logs/log_warning.log from file logs/log_warning.log with content encoding "gzip", mime type "text/plain" and expiry 2019-11-18T05:58:28.630Z
[taskcluster 2018-11-18T06:21:42.327Z] Uploading artifact public/test_info/marionette_errorsummary.log from file build/blobber_upload_dir/marionette_errorsummary.log with content encoding "gzip", mime type "text/plain" and expiry 2019-11-18T05:58:28.630Z
[taskcluster 2018-11-18T06:21:42.735Z] Uploading artifact public/test_info/marionette_raw.log from file build/blobber_upload_dir/marionette_raw.log with content encoding "gzip", mime type "text/plain" and expiry 2019-11-18T05:58:28.630Z
[taskcluster 2018-11-18T06:21:43.221Z] Uploading artifact public/test_info/report.html from file build/blobber_upload_dir/report.html with content encoding "gzip", mime type "text/html; charset=utf-8" and expiry 2019-11-18T05:58:28.630Z
[taskcluster 2018-11-18T06:21:43.598Z] Uploading artifact public/test_info/resource-usage.json from file build/blobber_upload_dir/resource-usage.json with content encoding "gzip", mime type "application/json" and expiry 2019-11-18T05:58:28.630Z
[taskcluster:error] exit status 2
Flags: needinfo?(mh+mozilla)
Backout by shindli@mozilla.com:
https://hg.mozilla.org/integration/autoland/rev/e432617f7098
Backed out 6 changesets for causing Mn mozprocess automation failures in '/Users/cltbld/tasks/task_1542517470/build/venv/bin/python
Status: RESOLVED → REOPENED
Resolution: FIXED → ---
Depends on: 1507948
Pushed by rgurzau@mozilla.com:
https://hg.mozilla.org/integration/autoland/rev/70e906a17ad3
Use MOZ_APP_VERSION instead of FIREFOX_VERSION. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/c62948fc14e6
Move FIREFOX_VERSION to python configure. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/f28ea3184e5c
Move MOZ_APP_VERSION{,_DISPLAY} to python configure. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/fd205107b0f4
Move js_option around to make it available earlier. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/83e6f0d47a80
Move MOZILLA_OFFICIAL to init.configure. r=nalexander
https://hg.mozilla.org/integration/autoland/rev/88ba9f723934
Derive the default update channel from the application display version. r=nalexander
Target Milestone: mozilla65 → ---
What changed between the backout and the relanding?
Flags: needinfo?(mh+mozilla) → needinfo?(rgurzau)
Backed out changeset 88ba9f723934 (bug 1502457) for causing bug 1507838 and bug 1508836.
https://hg.mozilla.org/integration/mozilla-inbound/rev/31766e8403ad17eb5b384c52750cbb49d1327e2e

Left the rest landed, as they were preparatory steps mostly moving stuff from old configure to moz.configure.
Status: RESOLVED → REOPENED
Resolution: FIXED → ---
As mentioned above, it turns out this caused problems. And looking around at abuses of MOZ_UPDATE_CHANNEL, it turns out there are multiple abuses this didn't solve anyways.
Kate, what are the specific builds that are targeted by this test below you added in bug 1492174?

> const IS_NIGHTLY_OR_UNBRANDED_BUILD = ["nightly", "default"].includes(AppConstants.MOZ_UPDATE_CHANNEL);

Obviously, nightlies, but what is meant by unbranded builds?
Flags: needinfo?(khudson)
Specifically, is it about local builds?
:glandium this was backed out together with another bug for apparently the same issue, it turns out at that time that this backout wasn't needed so it was relanded, only to be backed out again in the end.
Flags: needinfo?(rgurzau)
I think we included "default" in order to also include local builds, most usages have been removed as of https://bugzilla.mozilla.org/show_bug.cgi?id=1499110; I think it would be fine to change it to just nightly for the one remaining usage. Would you recommend doing that?
Flags: needinfo?(khudson)
Comment on attachment 9025264 [details]
Bug 1502457 - Move MOZ_APP_VERSION{,_DISPLAY} to python configure.

It seems this patch broke both Thunderbird and SeaMonkey versioning. version.txt is in the config dir of the project as in Firefox. Previously MOZ_APP_VERSION did reflect the version here when building either comm/mail or comm/suite. Now it is either the milestone or the Firefox browser version. Not sure which because both are equal in Nightly. 

Is there a way to instruct configure to use the correct version.txt file?
Flags: needinfo?(mh+mozilla)
version issue seems to be fixed in Bug 1515504
Flags: needinfo?(mh+mozilla)
Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.