I outlined the recent work regarding using ppp/adb networking to replace
wifi networking for Autophone's remote testing in yesterday's
Mobile/Testing meeting
<https://wiki.mozilla.org/Mobile/Testing/03_19_14#Autophone>.
I am backfilling mozilla-central and fx-team to 2013-03-13 and
mozilla-inbound to 2013-03-18. mozilla-central and fx-team will take a
couple of days to catch up though mozilla-inbound is current and keeping up.
The first results are coming in
<http://phonedash.mozilla.org/#/org.mozilla.fennec/throbberstop/remote-blank/norejected/2014-03-18/2014-03-20/notcached/errorbars/standarderror>
and I would like your opinion if they are valid.
It appears that the series of Promise.jsm landings caused a series of
regressions of fx-team which were ultimately fixed by the most recent
Promise.jsm landing. These regressions were only apparent on the remote
tests (http over adb/ppp) and not the local tests (content loaded from
the sdcard).
* regression #1 (major)
regression in throbberstop for remote tests only. also introduced noise.
<http://hg.mozilla.org/integration/fx-team/pushloghtml?fromchange=806f7c40cccf&tochange=af15bd900a2a>
candidates:
bug 928096 - UI for Tab streaming r=mfinkle
<https://bugzilla.mozilla.org/show_bug.cgi?id=928096>
Bug 984788 - Convert legacy uses of promise.js to Promise.jsm. rs=Yoric
<https://bugzilla.mozilla.org/show_bug.cgi?id=984788>
throbberstop gradually improved (though not 100%)
<http://hg.mozilla.org/integration/fx-team/pushloghtml?fromchange=af15bd900a2a&tochange=8b9f265aae5a>
* regression #2 (major)
regression in throbberstop for remote tests only.
<http://hg.mozilla.org/integration/fx-team/pushloghtml?fromchange=8b9f265aae5a&tochange=d16f3e084b27>
candidates:
Bug 818584 - Discard consecutive duplicate bookmark backups by comparing
hashes r=mano Original patch by Raymond Lee <raym...@appcoast.com> f=mak
<https://bugzilla.mozilla.org/show_bug.cgi?id=818584>
regression #2 and the noise appear to have been fixed as well as
most but not all of the remaining issue from regression #1 by the
backout of bug 818584:
<http://hg.mozilla.org/integration/fx-team/pushloghtml?fromchange=ad4b3715aef4&tochange=f5f7d7692869>
candidate:
<http://hg.mozilla.org/integration/fx-team/rev/f5f7d7692869>
Ryan VanderMeulen — Backed out changeset d16f3e084b27 (bug 818584) for
intermittent timeouts in test_bookmark_store.js. CLOSED TREE
* regression #3 (minor)
small regression in throbberstop for remote blank and slightly larger
regressionin throbberstop for remote twitter. It is difficult to tell
if the measurements actually signal a real regression.
<http://hg.mozilla.org/integration/fx-team/pushloghtml?fromchange=67a7d2c4e987&tochange=c7ee33d17226>
* regression #4 (major)
regression in throbberstop in remote.
<http://hg.mozilla.org/integration/fx-team/pushloghtml?fromchange=c7ee33d17226&tochange=a1945d5e465f>
candidates:
Bug 985318 - Convert to Promise.jsm in the Toolkit "content" folder. r=mak
<https://bugzilla.mozilla.org/show_bug.cgi?id=985318>
Bug 985523 - Convert to Promise.jsm in Places. r=mak
<https://bugzilla.mozilla.org/show_bug.cgi?id=985523>
All of the remaining regressions appear to be fixed by
<http://hg.mozilla.org/integration/fx-team/pushloghtml?fromchange=a1945d5e465f&tochange=d4cb68734521>
candidate:
Bug 984809 - Remove deprecated promise.js usage in B2G. r=ochameau
<https://bugzilla.mozilla.org/show_bug.cgi?id=984809>
Do these results make sense? Would you have found them useful?
/bc
_______________________________________________
mobile-firefox-dev mailing list
mobile-firefox-dev@mozilla.org
https://mail.mozilla.org/listinfo/mobile-firefox-dev
