1209312 - Bring back xpcshell support for |./mach test|

Status: RESOLVED FIXED

(Testing :: XPCShell Harness, defect)

Description

[Chris Manchester (limited bugmail, email directly)]

The test_objects stuff is gone.

Comment 1

[Chris Manchester (limited bugmail, email directly)]

Attachment: MozReview Request: Bug 1209312 - Accept test_objects in testing/xpcshell/mach_commands.py to support |./mach test|. r=jgraham

Bug 1209312 - Accept test_objects in testing/xpcshell/mach_commands.py to support |./mach test|. r=jgraham

Comment 2

[James Graham [:jgraham]]

Comment on attachment 8666989 [details]
MozReview Request: Bug 1209312 - Accept test_objects in testing/xpcshell/mach_commands.py to support |./mach test|. r=jgraham

https://reviewboard.mozilla.org/r/20631/#review18533

::: testing/xpcshell/mach_commands.py:343
(Diff revision 1)
> +            m = TestManifest()

In general I'm not convinced about this pattern where we construct a TestManifest object in mach rather than just passing the path in and letting the harness deal with it. I thought we had decided we wanted mach to be a thin wrapper that just set arguments based on the local environment.

But not a blocker to landing this if you think this design has some advantage.

Comment 3

[Chris Manchester (limited bugmail, email directly)]

https://reviewboard.mozilla.org/r/20631/#review18533

> In general I'm not convinced about this pattern where we construct a TestManifest object in mach rather than just passing the path in and letting the harness deal with it. I thought we had decided we wanted mach to be a thin wrapper that just set arguments based on the local environment.
> 
> But not a blocker to landing this if you think this design has some advantage.

I'm not particularly fond of the design, but it's needed for mach test. This is just fixing a regression.

Comment 4

[James Graham [:jgraham]]

https://reviewboard.mozilla.org/r/20631/#review18533

> I'm not particularly fond of the design, but it's needed for mach test. This is just fixing a regression.

It isn't really neded for mach test, you can just pull out the manifest path from the data in test_objects and pass that through to the harness, rather than first constructing a manifest. Anyway, like I say you don't have to change this since it's a pattern that's used elsewhere. I just think it's not better than the alternative.

Comment 5

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/8b3a4fdebdfc

Comment 6

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/8b3a4fdebdfc