1256290 - test_crash_manager.jsm fails locally because of a round-trip failure in the environment (GPUActive: undefined)

Status: RESOLVED FIXED

(Toolkit :: Telemetry, defect, P3)

Description

[Benjamin Smedberg]

On my local Windows machine, test_crash_manager.jsm always fails at:

Assert.deepEqual(found.environment, theEnvironment, "The saved environment should be present");

I dumped these objects and the difference is a property  system.gfx.adapters[1].GPUActive === undefined

Since undefined props disappear when round-tripped through JSON, the serialized version doesn't exactly match.

The code path that gets this value is:
http://hg.mozilla.org/mozilla-central/annotate/f0c0480732d3/toolkit/components/telemetry/TelemetryEnvironment.jsm#l1319

Which pulls a property off nsIGfxInfo but I'm pretty sure there just shouldn't be an extra space in the prop name at 1319. I'm going to fix getGfxField to be less tolerant of undefined values.

Comment 1

[Benjamin Smedberg]

Attachment: MozReview Request: Bug 1256290 - isGPUActive for the second graphics adapter doesn't work correctly, and this also means we should tighten up getGfxField a little bit r?gfritzsche

Review commit: https://reviewboard.mozilla.org/r/39761/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/39761/

Comment 2

[Benjamin Smedberg]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=c9c545ecf09b

Comment 3

[Georg Fritzsche [:gfritzsche]]

Comment on attachment 8730226 [details]
MozReview Request: Bug 1256290 - isGPUActive for the second graphics adapter doesn't work correctly, and this also means we should tighten up getGfxField a little bit r?gfritzsche

https://reviewboard.mozilla.org/r/39761/#review36553

Thanks.

Comment 4

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/dce690007f02

Comment 5

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/dce690007f02