895940 - Mirror mozbase -> m-c for bug 865349 + others

Status: RESOLVED FIXED

(Testing :: Mozbase, defect)

Description

[Andrew Halberstadt [:ahal]]

I need at least mozprofile, mozdevice and mozrunner mirrored to m-c for bug 865349. Joel also wants mozinfo. I wonder if I should just do a full sync. I imagine some of the lesser used modules are getting pretty out of date by now.

Comment 1

[Andrew Halberstadt [:ahal]]

I pushed a full sync of mozbase to try just to see what would happen:
https://tbpl.mozilla.org/?tree=Try&rev=5fe83b2d65ca

This also adds most of the new tests that ffledgling has been working on. I still need to do version bumps where appropriate (assuming try run works).

Comment 2

[Andrew Halberstadt [:ahal]]

Looks like mozinfo tests are failing:

======================================================================
FAIL: test_find_and_update_file (test.TestMozinfo)
Test that mozinfo.find_and_update_from_json can
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/builds/slave/try-osx64_g-000000000000000000/build/testing/mozbase/mozinfo/tests/test.py", line 56, in test_find_and_update_file
    self.assertEqual(mozinfo.find_and_update_from_json(self.tempdir), j)
AssertionError: u'/builds/slave/try-osx64_g-000000000000000000/build/obj-firefox/mozinfo.json' != '/var/folders/h5/8sbn_pwn5zz1_vknq9c_vxg800000w/T/tmpgNI2Jq/mozinfo.json'

and 

======================================================================
FAIL: test_find_and_update_file_invalid_json (test.TestMozinfo)
Test that mozinfo.find_and_update_from_json can
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/builds/slave/try-osx64_g-000000000000000000/build/testing/mozbase/mozinfo/tests/test.py", line 65, in test_find_and_update_file_invalid_json
    self.assertRaises(ValueError, mozinfo.find_and_update_from_json, self.tempdir)
AssertionError: ValueError not raised

Comment 3

[Andrew Halberstadt [:ahal]]

The above errors happen because mozinfo.find_and_update_from_json() ignores the 'dirs' parameter completely if the mozbuild module is detected. Filed bug 896038 to take care of it.

Comment 4

[Jeff Hammel]

Is this proposing to version bump all packages?  IMHO, we should do so if we're going to sync the whole tree

Comment 5

[Andrew Halberstadt [:ahal]]

Yeah, well at least the ones that have changes not in m-c.

Comment 6

[Andrew Halberstadt [:ahal]]

Try run with fix for mozinfo tests:
https://tbpl.mozilla.org/?tree=Try&rev=584a1889981c

Comment 7

[Andrew Halberstadt [:ahal]]

Attachment: Patch 1.0 - bump mozbase modules that need it

This patch bumps all the mozbase modules that had changes after the last version bump. Modules that were bumped in their last commit, or only had test changes since, were left alone.

I've done all the appropriate tags in my local branch. I've also updated (the way out of date) mozilla-central tag. I won't mirror any changes that come in at any point after now.

Comment 8

[Andrew Halberstadt [:ahal]]

Attachment: Patch 2.0 - update mozbase in m-c

This is what's going into m-c, including the changes from the previous patch. Let me know if you are busy, and I can reassign.

Comment 9

[Andrew Halberstadt [:ahal]]

Comment on attachment 779211 [details] [diff] [review]
Patch 1.0 - bump mozbase modules that need it

Jeff appears to be out, got an irc r+ from wlach

Comment 10

[Andrew Halberstadt [:ahal]]

Attachment: Patch 2.0 - update mozbase in m-c

Just noticed I had an out of date patch uploaded

Comment 11

[Jeff Hammel]

(In reply to Andrew Halberstadt [:ahal] from comment #7)
<snip/>
> I've done all the appropriate tags in my local branch. I've also updated
> (the way out of date) mozilla-central tag. I won't mirror any changes that
> come in at any point after now.

I thought we decided not to update the mozilla-central tag when we moved to a policy of updating individual packages (fall 2012 workweek).  I've actually tried to delete this tag (amongst others) multiple times, but ABICT it gets pushed from someones' local branch and since its annoying and misleading but hasn't been a problem, I haven't ticketed the need for this tag or others that shouldn't be in mozilla/mozbase to be erased.

Comment 12

[Jeff Hammel]

(In reply to Andrew Halberstadt [:ahal] from comment #9)
> Comment on attachment 779211 [details] [diff] [review]
> Patch 1.0 - bump mozbase modules that need it
> 
> Jeff appears to be out, got an irc r+ from wlach

My apologies.  I've been out sick today and wasn't doing too well yesterday either.  Feeling a bit better now so hopefully will be back in the game tomorrow.

I saw the mozbase side of things was already pushed \o/

Is all that is left landing on m-i/m-c ?

Comment 13

[Andrew Halberstadt [:ahal]]

For module bump:
https://github.com/mozilla/mozbase/commit/c7417faafd0238e2a8da1a1cb08195caff44fde7

(In reply to Jeff Hammel [:jhammel] from comment #12)
> I saw the mozbase side of things was already pushed \o/
> 
> Is all that is left landing on m-i/m-c ?

Yep! It's ready to go with a green try run:
https://tbpl.mozilla.org/?tree=Try&rev=75291fe91e09

The initial errors were because of missing modules in the mozharness script and have been fixed.

Comment 14

[Andrew Halberstadt [:ahal]]

Comment on attachment 779918 [details] [diff] [review]
Patch 2.0 - update mozbase in m-c

Jeff, want to give it a quick once over? Will and I were discussing whether mirrors that pass try runs really need formal reviews or not. Guess it doesn't hurt to be safe.

Comment 15

[Andrew Halberstadt [:ahal]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/a13dafd65d1c

Comment 16

[Mike Hommey [:glandium]]

Backed out because it broke PGO builds:
https://hg.mozilla.org/integration/mozilla-inbound/rev/7c78c3a98313

https://tbpl.mozilla.org/php/getParsedLog.php?id=25698383&tree=Mozilla-Inbound
Traceback (most recent call last):
  File "/builds/slave/m-in-lx-pgo-000000000000000000/build/build/pgo/profileserver.py", line 66, in <module>
    env=env)
  File "/builds/slave/m-in-lx-pgo-000000000000000000/build/testing/mozbase/mozrunner/mozrunner/local.py", line 214, in __init__
    Runner.__init__(self, profile, binary, **kwargs)
TypeError: __init__() got an unexpected keyword argument 'cmdargs'
Exception AttributeError: "'FirefoxRunner' object has no attribute 'process_handler'" in <bound method FirefoxRunner.cleanup of <mozrunner.local.FirefoxRunner object at 0x2f53b90>> ignored

Comment 17

[Andrew Halberstadt [:ahal]]

Attachment: Patch 3.0 - updated with fix from bug 897967

This is the same as the previous patch except with the fix from bug 897967 as well. Carrying r+ forward.

Comment 18

[Andrew Halberstadt [:ahal]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/4f7c7217108a

Comment 19

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/4f7c7217108a

Comment 20

[Martijn Wargers (dead)]

I updated my b2g build and I'm now getting this error while trying to start up b2g mochitest emulator: 
  File "/home/mwargers/B2G/objdir-gecko/_tests/testing/mochitest/runtests.py", line 178, in __init__
    mozinfo.find_and_update_from_json(SCRIPT_DIR)
AttributeError: 'module' object has no attribute 'find_and_update_from_json'

Perhaps, I need to update some module or something?

Comment 21

[Andrew Halberstadt [:ahal]]

Yep, try 'pip install --upgrade mozinfo', or better yet create a new virtualenv (https://pypi.python.org/pypi/virtualenv) and install an up to date copy of mozbase into it.

For the record I'm working on getting mach target support for b2g in bug 867817 so developers don't have to worry about problems like this anymore.