990639 - Add support for e10s tests to TBPL

Status: RESOLVED FIXED

(Tree Management Graveyard :: TBPL, defect)

Description

[Ryan VanderMeulen [:RyanVM]]

See bug 989533 comment 11 and down.

Comment 1

[Ryan VanderMeulen [:RyanVM]]

I have the basic chunking working on this. Where things get complicated is modifying TBPL to only show the mochitest chunk numbers for the like they do for the non-e10s jobs.

Comment 2

[Ryan VanderMeulen [:RyanVM]]

OK, I've got mochitests working. The reftest-e10s group is going to be a bit ugly in that it's going to look like R-e10s(C J R-e10s) since AFAIK I can't split the group symbol from the job symbol.

Comment 3

[Ryan VanderMeulen [:RyanVM]]

Attachment: add new e10s groups

This works for as much testing as I've been able to do (the jobs that ran for awhile today on TBPL look right anyway). Not happy about the aforementioned reftest symbols issue, but I'm not sure of a good way around it either.

Comment 4

[Ed Morley [:emorley]]

Comment on attachment 8400148 [details] [diff] [review]
add new e10s groups

Review of attachment 8400148 [details] [diff] [review]:
-----------------------------------------------------------------

Looks good :-)

Let's land this for now and we can always land a followup to _machineResultLink() to handle the !Mochitest case more nicely.

Comment 5

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/webtools/tbpl/rev/72d11b38beca

(In reply to Ed Morley [:edmorley UTC+0] from comment #4)
> Let's land this for now and we can always land a followup to
> _machineResultLink() to handle the !Mochitest case more nicely.

You're thinking of things like handling it better for cases like Android 2.3 xpcshell?

Also, what's your take on the reftest symbols? Any ideas for how we could display that better?

Comment 6

[Ed Morley [:emorley]]

Changes to _machineResultLink() would fix the reftest symbols, not the former.

Comment 7

[Ryan VanderMeulen [:RyanVM]]

Ah, OK.

Comment 8

[Ryan VanderMeulen [:RyanVM]]

In production.