1388417 - 1.85% installer size (android-4-0-armv7-api15) regression on push 6df7dff00b94cf7232b7d507eee373fe6b0435f0 (Tue Aug 8 2017)

Status: RESOLVED FIXED

(Firefox for Android Graveyard :: General, defect, P1)

Description

[William Lachance (:wlach)]

Rather large increase in the .apk size from the new photon resources. Is there anything we can do to reduce/mitigate?

--

We have detected a build metrics regression from push:

https://hg.mozilla.org/integration/autoland/pushloghtml?changeset=6df7dff00b94cf7232b7d507eee373fe6b0435f0

As author of one of the patches included in that push, we need your help to address this regression.

Regressions:

  2%  installer size summary android-4-0-armv7-api15 opt      36,309,723.83 -> 36,980,326.50


You can find links to graphs and comparison views for each of the above tests at: https://treeherder.mozilla.org/perf.html#/alerts?id=8631

On the page above you can see an alert for each affected platform as well as a link to a graph showing the history of scores for this test. There is also a link to a treeherder page showing the jobs in a pushlog format.

To learn more about the regressing test(s), please see: https://developer.mozilla.org/en-US/docs/Mozilla/Performance/Automated_Performance_Testing_and_Sheriffing/Build_Metrics

Comment 1

[Jing-wei Wu [:jwu]]

The root cause is that we've added some new images/icons in Nightly for new visual refresh, aka Photon. There are few things we can do to reduce the APK size:

1. Remove resources that are unused in new visual design, we use bug 1375351 to track this issue.
2. Compress images/icons, or maybe use vector drawable instead.

Comment 2

[Jing-wei Wu [:jwu]]

Hi :wlach,

Now Fennec apk size has been reduced to around 36,457,239. Can we close this bug now?

https://treeherder.mozilla.org/perf.html#/graphs?series=autoland,1545789,1,2

Comment 3

[William Lachance (:wlach)]

This isn't a talos regression, but we can probably use the policy here as a guideline https://wiki.mozilla.org/Buildbot/Talos/RegressionBugsHandling

Is the improvement due to fixing the root cause in comment 1, or due to other, unrelated work? 

* If there are still mitigations for this regression that are possible and make sense to implement, we should probably keep this open (along with outlining what needs to be fixed). 
* If your recommendations in comment 1 were implemented and that's responsible for the improvement, then we can resolve as fixed.
* If there are no mitigations that make sense to implement and the regression is just something we have to accept, I would mark this as wontfix.

Comment 4

[Jing-wei Wu [:jwu]]

We do still have some space to reduce the APK size and there is a meta bug 942609 that records all the related works.

Comment 5

[William Lachance (:wlach)]

(In reply to Jing-wei Wu [:jwu] from comment #4)
> We do still have some space to reduce the APK size and there is a meta bug
> 942609 that records all the related works.

Argh, I didn't actually mean to mark this as fixed. But if the issues here are being dealt with elsewhere, then let's just leave this be.