Closed
Bug 524125
Opened 15 years ago
Closed 15 years ago
Provide makefile target for bloat tests and don't hard-code path to symbols
Categories
(MailNews Core :: Build Config, defect)
MailNews Core
Build Config
Tracking
(thunderbird3.0 .1-fixed)
VERIFIED
FIXED
Thunderbird 3.1a1
Tracking | Status | |
---|---|---|
thunderbird3.0 | --- | .1-fixed |
People
(Reporter: standard8, Assigned: standard8)
References
(Blocks 1 open bug)
Details
(Keywords: fixed-seamonkey2.0.3)
Attachments
(3 files)
(deleted),
patch
|
gozer
:
review+
standard8
:
approval-thunderbird3.0.1+
|
Details | Diff | Splinter Review |
(deleted),
patch
|
gozer
:
review+
standard8
:
approval-thunderbird3.0.1+
|
Details | Diff | Splinter Review |
(deleted),
patch
|
Details | Diff | Splinter Review |
Bug 523330 removed automation.SYMBOLS_PATH and in the process killed our bloat tests.
Reading the comments on that bug, I guess the best way to deal with this would be to create our own makefile target the bloat test.
I think my short term fix would be to disable symbols or maybe hard-code (as it is only really tinderboxes going to be using it), and then the longer term fix would be to actually create the makefile target.
Comment 1•15 years ago
|
||
Sorry, I didn't know you had any independent scripts using automation.py. A makefile target will probably make things easier in the long run, since you can easily change things without having to change build automation. (Of course, with our packaged tests, we're right back where we started.)
Assignee | ||
Comment 2•15 years ago
|
||
This literally hard-codes the path (thankfully bug 522115 comment 14 told me what this was) until we can get some time to do a better fix.
Attachment #408060 -
Flags: review?(gozer)
Assignee | ||
Comment 3•15 years ago
|
||
(In reply to comment #2)
> This literally hard-codes the path (thankfully bug 522115 comment 14 told me
> what this was) until we can get some time to do a better fix.
(this is a trunk-only fix)
(In reply to comment #1)
> Sorry, I didn't know you had any independent scripts using automation.py.
Yep, bloat uses it and our mozmill tests use it. iirc neither use all the functionality, but they both use bits of it in some way.
Comment 4•15 years ago
|
||
Comment on attachment 408060 [details] [diff] [review]
[checked in] Band-aid runtest.py
Works for me, let's make sure to remove it once bug 524125 is taken care of.
Attachment #408060 -
Flags: review?(gozer) → review+
Assignee | ||
Comment 5•15 years ago
|
||
Comment on attachment 408060 [details] [diff] [review]
[checked in] Band-aid runtest.py
Checked in: http://hg.mozilla.org/comm-central/rev/47a01c703072
Attachment #408060 -
Attachment description: Band-aid runtest.py → [checked in] Band-aid runtest.py
Assignee | ||
Comment 6•15 years ago
|
||
Morphing bug into the next step.
Summary: Bloat tests busted by bug 523330 - automation.SYMBOLS_PATH does not exist → Provide makefile target for bloat tests and don't hard-code path to symbols
Assignee | ||
Comment 7•15 years ago
|
||
This reworks the existing way to run the bloat tests to provide a "make mailbloat" target from the top of the objdir.
This seems to work fine on Mac, and also adds some tidy up to remove the files installed into chrome at the end of the test (really it should be an extension or something, but I'm still half hoping to replace this with mozmill at some stage).
Assignee | ||
Comment 8•15 years ago
|
||
The buildbot changes. This swaps buildbot to "make mailbloat". I've also pulled across some of the settings from the core buildbotcustom leak test, including the upload/download of logs and aligning some of the warnOnFailure/haltOnFailure settings.
This obviously could do with some testing especially as I've assumed what will work in some places.
Attachment #411089 -
Flags: review?(gozer)
Updated•15 years ago
|
Attachment #411088 -
Flags: review?(gozer) → review+
Comment 9•15 years ago
|
||
(In reply to comment #8)
> Created an attachment (id=411089) [details]
> buildbot changes
>
> The buildbot changes. This swaps buildbot to "make mailbloat". I've also pulled
> across some of the settings from the core buildbotcustom leak test, including
> the upload/download of logs and aligning some of the
> warnOnFailure/haltOnFailure settings.
Yes, looks like a good idea in general, I'll try and test this patch in staging shortly. I'd love to switch over to CC*Factories for bloat stuff, but last time I tried, there were still quite a few borked things. This is a step in the right direction.
> This obviously could do with some testing especially as I've assumed what will
> work in some places.
I'll report back on the results from staging
Comment 10•15 years ago
|
||
Comment on attachment 411089 [details] [diff] [review]
buildbot changes
patch applied in staging and encountered 2 issues.
had to s/self.env/env/g, easy fix
the various STAGE_* settings are not visible in mozillaleak.py, and should be passed in via the branch object argument. Unfortunately, these settings are now set in most branches, and rely on branch.get('STAGE_*', STAGE_*) type of logic.
That's wrong logic, anyhow, these should be determined upfront and stashed in the branch configuration. I'll get that fixed.
Assignee | ||
Comment 11•15 years ago
|
||
Assigning to gozer as he needs to do the buildbot work.
Assignee: bugzilla → gozer
Assignee | ||
Comment 12•15 years ago
|
||
Comment on attachment 408060 [details] [diff] [review]
[checked in] Band-aid runtest.py
I'm going to be landing this on the 1.9.1 branch as it is test-only code, and something we need to support buildbot better. a=Standard8
Attachment #408060 -
Flags: approval-thunderbird3.0.1+
Assignee | ||
Updated•15 years ago
|
Attachment #411088 -
Flags: approval-thunderbird3.0.1+
Assignee | ||
Comment 13•15 years ago
|
||
comm-* changes Checked in:
http://hg.mozilla.org/comm-central/rev/9911eaa87f34
http://hg.mozilla.org/releases/comm-1.9.1/rev/aca404d1fcdf
Assignee | ||
Updated•15 years ago
|
Attachment #411088 -
Attachment description: Add mailbloat target → [checked in] Add mailbloat target
Assignee | ||
Comment 14•15 years ago
|
||
Update:
make mailbloat seems to be working fine on all platform.
Builders suffering from not being able to upload the logs, and hence can't download them either. Note that the two upload failures are slightly different.
Comment 15•15 years ago
|
||
Issues with staging server name and path names, now corrected, working perfectly.
Status: ASSIGNED → RESOLVED
Closed: 15 years ago
Resolution: --- → FIXED
Assignee | ||
Comment 16•15 years ago
|
||
Comment on attachment 411089 [details] [diff] [review]
buildbot changes
Clearing now obsolete review request.
Attachment #411089 -
Flags: review?(gozer)
Assignee | ||
Comment 17•15 years ago
|
||
Marking as fixed-1.9.1, although this change was test code only.
status-thunderbird3.0:
--- → .1-fixed
Flags: in-testsuite-
Assignee | ||
Comment 18•15 years ago
|
||
Verifying as fixed as the bloat boxes are running nicely with this test-only change.
Status: RESOLVED → VERIFIED
Keywords: verified-thunderbird3.0
Target Milestone: --- → Thunderbird 3.1a1
Updated•15 years ago
|
Keywords: fixed-seamonkey2.0.3
Updated•13 years ago
|
Assignee: gozer → mbanner
You need to log in
before you can comment on or make changes to this bug.
Description
•