Make SeaMonkey buildbot configs base on abstracted buildbotcustom factories

RESOLVED FIXED in seamonkey2.0b1

Status

defect
RESOLVED FIXED
10 years ago
9 years ago

People

(Reporter: kairo, Assigned: kairo)

Tracking

Trunk
seamonkey2.0b1
Dependency tree / graph
Bug Flags:
in-testsuite -

Firefox Tracking Flags

(Not tracked)

Details

Attachments

(2 attachments, 2 obsolete attachments)

The SeaMonkey buildbot configs are currently using custom steps, we should convert them to the generalized steps in buildbotcustom, see also bug 478229.
Blocks: 485821
Along with getting the bug 485436 fix to actually work, I investigated running the generic UnittestBuildFactory from buildbotcustom and found some still-existing problems there, but converted our test buildbot configs to use almost the same steps as we have in use there.
Remove (comment 1) workaround now that bug 479225 patch Dv1a is checked in.

While there, "fix" HGURL story.
Attachment #371592 - Flags: review?(kairo)
Depends on: 479225
Posted patch (Bv1) Clean up mozconfig (obsolete) — Splinter Review
All 3 platforms:
*Update URL.
*Remove obsolete (commented out) --enable-mochitest.

Windows:
*Remove commented out --disable-installer.
*Add -s.
Attachment #371596 - Flags: review?(kairo)
(In reply to comment #3)
> *Add -s.

Why that? do "mozilla2" configs for Firefox have it as well? I want to be/stay as near as possible to those configs.
Comment on attachment 371592 [details] [diff] [review]
(Av1) Use leakThreshold directly now

>-        if mochitest_leak_threshold:
>-            self.mochitest_leak_threshold = mochitest_leak_threshold
>-        if mochichrome_leak_threshold:
>-            self.mochichrome_leak_threshold = mochichrome_leak_threshold
>-        if mochibrowser_leak_threshold:
>-            self.mochibrowser_leak_threshold = mochibrowser_leak_threshold

Please leave those in, we do this for all arguments when calling such classes everywhere in buildbotcustom.

>          leakThreshold=mochitest_leak_threshold,

>+         leakThreshold=mochichrome_leak_threshold,

>+         leakThreshold=mochibrowser_leak_threshold,

Make those use self.*

>-# Since they're both right here it shouldn't be a problem to keep them in sync.
>-HGURL = 'http://hg.mozilla.org/'
> HGHOST = 'hg.mozilla.org'
>+HGURL = 'http://%s/' % HGHOST

Please don't touch this at this time, unless the "mozilla2" configs for Firefox have the same change.
Attachment #371592 - Flags: review?(kairo) → review-
Depends on: 488827
Av1, with comment 5 suggestion(s).
Attachment #371592 - Attachment is obsolete: true
Attachment #373295 - Flags: review?(kairo)
Bv1, with comment 4 suggestion(s).


(In reply to comment #4)
> (In reply to comment #3)
> > *Add -s.
> 
> Why that?

Because Linux and MacOSX had it.

> do "mozilla2" configs for Firefox have it as well?

No, none, see
http://mxr.mozilla.org/build/search?string=MOZ_MAKE_FLAGS.*-s&regexp=on&case=on&find=%2Fmozconfig

> I want to be/stay as near as possible to those configs.

Then let's remove these two.
Attachment #371596 - Attachment is obsolete: true
Attachment #373298 - Flags: review?(kairo)
Attachment #371596 - Flags: review?(kairo)
Comment on attachment 373298 [details] [diff] [review]
(Bv1a) Clean up mozconfig
[Checkin: See comment 10]

>@@ -1,14 +1,13 @@
>-# http://hg.mozilla.org/build/buildbot-configs/seamonkey-unittest/mozconfig-centos
>+# http://hg.mozilla.org/build/buildbot-configs/file/tip/seamonkey-unittest/linux/comm-central/unittest/mozconfig

Please just go an remove that line and the and the empty one under it from all the files.

r=me with that (feel free to check in this one).

I still dislike misusing this bug report for not-really-related patches, though.
Attachment #373298 - Flags: review?(kairo) → review+
Comment on attachment 373295 [details] [diff] [review]
(Av2) Use leakThreshold directly now
[Checkin: comment 9+12+13]

r=me but please don't check this in, I'll incorporate it in the patch/checkins for bug 488116.
Attachment #373295 - Flags: review?(kairo) → review+
Comment on attachment 373298 [details] [diff] [review]
(Bv1a) Clean up mozconfig
[Checkin: See comment 10]


http://hg.mozilla.org/build/buildbot-configs/rev/187e0971a889
Attachment #373298 - Attachment description: (Bv1a) Clean up mozconfig → (Bv1a) Clean up mozconfig [Checkin: Comment 10]
Comment on attachment 373298 [details] [diff] [review]
(Bv1a) Clean up mozconfig
[Checkin: See comment 10]


(In reply to comment #10)
> http://hg.mozilla.org/build/buildbot-configs/rev/187e0971a889

Bv1a, with comment 8 suggestion(s).
Attachment #373298 - Attachment description: (Bv1a) Clean up mozconfig [Checkin: Comment 10] → (Bv1a) Clean up mozconfig [Checkin: See comment 10]
Depends on: 488116
http://hg.mozilla.org/build/buildbot-configs/rev/a9cdae769da3 landed to use buildbotcustom CCUnittestBuildFactory as introduced in bug 488116, including the changes to config.py from attachment 373295 [details] [diff] [review]. I'll activate this on the buildbots tomorrow, probably.
Comment on attachment 373295 [details] [diff] [review]
(Av2) Use leakThreshold directly now
[Checkin: comment 9+12+13]

I'm moving the remaining parts to bug 489478.
Attachment #373295 - Attachment description: (Av2) Use leakThreshold directly now → (Av2) Use leakThreshold directly now [Checkin: comment 9+12+13]
CCUnittestBuildFactory is deployed on SeaMonkey buildbots now.
Depends on: 486540
Summary: Make SeaMonkey buildbot configs base on abstracted buildbotcustom steps → Make SeaMonkey buildbot configs base on abstracted buildbotcustom factories
Blocks: 492221
Depends on: 464707
Depends on: 492265
I have a testing configuration of the new SeaMonkey buildbot configs reporting to the SeaMonkey-Ports waterfall page and the configs themselves in http://hg.mozilla.org/users/kairo_kairo.at/bbconf-sm2/ for testing, will go into buildbot-configs when ready for production.

Not everything is turned on yet and some known reds are there, will turn on further features step by step as showstoppers are fixed, up to a point where we can switch the new config to production :)
Depends on: 492253
Depends on: 464325
The new configs build upon the abstracted factories are running in testing mode now, I'll use bug 485821 for moving to production as sucking the old VMs into the new pools and moving to production is basically the same step.
Assignee: nobody → kairo
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → FIXED
Target Milestone: --- → seamonkey2.0b1
Blocks: 493448
Blocks: 493451
Flags: in-testsuite-
Component: Project Organization → Release Engineering
QA Contact: organization → release
You need to log in before you can comment on or make changes to this bug.