Configure Qt API using config file #130

nicoddemus · 2016-05-12T23:22:11Z

The big work here was of course to implement lazy-loading of Qt classes from qt_compat.

The-Compiler · 2016-05-13T04:53:06Z

Oh wow, crazy stuff! 😆 👍

What do you think about setting a flag when pytest_configure was executed (or maybe pytest has already one?), and then checking that when doing the actual Qt import (to make sure no import happened before the config was read)?

As for testing, I guess you could set up an environment with multiple Qt libraries installed so pytest-qt would pick the wrong one, then run in a subprocess (to make sure nothing is imported) with the config option, and e.g. look at the version output (or simply run a test in the subprocess) to make sure the right Qt API was picked?

nicoddemus · 2016-05-13T12:21:24Z

What do you think about setting a flag when pytest_configure was executed (or maybe pytest has already one?), and then checking that when doing the actual Qt import (to make sure no import happened before the config was read)?

That does not seem necessary: all imports will happen in the set_qt_api method, and that method is called in pytest_configure already. Or do you see some scenario or that might not be True?

As for testing, I guess you could set up an environment with multiple Qt libraries installed so pytest-qt would pick the wrong one, then run in a subprocess (to make sure nothing is imported) with the config option, and e.g. look at the version output (or simply run a test in the subprocess) to make sure the right Qt API was picked?

Good idea, although I it will complicate the tox setup. But you gave me an idea: setting the ini variable to the wrong api should raise an appropriate ImportError when the plugin loads. So, in a PySide environment this ini file:

[pytest]
qt_api=pyqt4

Will fail make pytest fail with an ImportError: PyQt4, thus proving the variable is working. What do you think?

qt_compat

This change will break backward compatibility for the qt_compat module, since from pytestqt.qt_compat import QApplication (for example) will not work anymore. I think this will have to warrant a 2.0 release to follow semantic versioning consistently, even thought I don't think it will break much code out there. I will probably rename it to _qt_compat to remove it from the public API as well. Thoughts?

The-Compiler · 2016-05-13T14:18:38Z

What do you think about setting a flag when pytest_configure was executed (or maybe pytest has already one?), and then checking that when doing the actual Qt import (to make sure no import happened before the config was read)?

That does not seem necessary: all imports will happen in the set_qt_api method, and that method is called in pytest_configure already. Or do you see some scenario or that might not be True?

You're right! I thought using qt_api.QtFoo would automatically import it no matter what, but I didn't take a close enough look.

Though a test which makes sure importing before pytest_configure fails would be nice.

As for testing, I guess you could set up an environment with multiple Qt libraries installed so pytest-qt would pick the wrong one, then run in a subprocess (to make sure nothing is imported) with the config option, and e.g. look at the version output (or simply run a test in the subprocess) to make sure the right Qt API was picked?

Good idea, although I it will complicate the tox setup. But you gave me an idea: setting the ini variable to the wrong api should raise an appropriate ImportError when the plugin loads. So, in a PySide environment this ini file:
[pytest]
qt_api=pyqt4
Will fail make pytest fail with an ImportError: PyQt4, thus proving the variable is working. What do you think?

Sounds good to me.

qt_compat

This change will break backward compatibility for the qt_compat module, since from pytestqt.qt_compat import QApplication (for example) will not work anymore. I think this will have to warrant a 2.0 release to follow semantic versioning consistently, even thought I don't think it will break much code out there. I will probably rename it to _qt_compat to remove it from the public API as well. Thoughts?

Agreed. Is there anything else which would go into a 2.0 release? What about delaying 2.0 until after the pytest sprint? Then we can discuss if there's anything else we'd like to change which would need an API break (for example, I'd like to discuss making qt_wait_signal_raising default to True).

I also thought about if we should use/contribute to one of the existing wrappers instead, like pyqode.qt - but probably it's not worth the trouble, as I'd guess @ColinDuquesnoy is more interested in the needs of pyQode (for which it was written for) 😉

nicoddemus · 2016-05-14T12:52:19Z

Agreed. Is there anything else which would go into a 2.0 release? What about delaying 2.0 until after the pytest sprint?

Sure, sounds good!

for example, I'd like to discuss making qt_wait_signal_raising default to True

No discussion there! 😉 In fact it was supposed to change even before that, as stated in this note. But I agree that changing it into a 2.0 is better specially if it is not far away.

I also thought about if we should use/contribute to one of the existing wrappers instead, like pyqode.qt - but probably it's not worth the trouble

I also don't think it would be worth it: pytest-qt uses a very limited set of the Qt api that just our humble qt_compat module works well, I think (although a little ugly), and I wouldn't like to add an extra dependency without a very compelling reason.

ColinDuquesnoy · 2016-05-14T13:17:27Z

I also thought about if we should use/contribute to one of the existing wrappers instead, like pyqode.qt - but probably it's not worth the trouble, as I'd guess @ColinDuquesnoy is more interested in the needs of pyQode (for which it was written for) 😉

I also think it wouldn't be worth it. There are already many wrappers, nearly any project that wants to support pyside, pyqt4 and pyqt5 have their own wrappers, written for their own needs.

nicoddemus · 2016-06-29T23:52:56Z

tests/test_basics.py

+
+
+@pytest.mark.parametrize('option_api', ['pyqt4', 'pyqt4v2', 'pyqt5', 'pyside'])
+def test_qt_api_ini_config(testdir, option_api):


This is how I ended up testing the qt_api variable in the end

coveralls · 2016-06-30T00:04:58Z

Coverage decreased (-6.8%) to 93.179% when pulling 020b1b5 on qt-api-ini into c8edfa2 on master.

coveralls · 2016-06-30T00:04:59Z

Coverage decreased (-6.8%) to 93.179% when pulling 020b1b5 on qt-api-ini into c8edfa2 on master.

coveralls · 2016-06-30T00:09:11Z

Coverage decreased (-4.8%) to 95.16% when pulling 1cb63ee on qt-api-ini into c8edfa2 on master.

coveralls · 2016-06-30T00:21:07Z

Coverage decreased (-2.0%) to 98.039% when pulling 49cf8e6 on qt-api-ini into c8edfa2 on master.

Fix #129

nicoddemus · 2016-06-30T09:13:00Z

@The-Compiler rebased and I think this is ready for review. Could you take a look when you have the time please? 😁

The-Compiler · 2016-06-30T09:21:04Z

CHANGELOG.rst

+* Now which Qt binding ``pytest-qt`` should use can be configured by the ``qt_api`` config option.
+  Thanks `@The-Compiler`_ for the request (`#129`_).
+
+- ``PYTEST_QT_FORCE_PYQT`` environment variable is no longer supported.


Shouldn't this be under "breaking changes"?

Oops, bad merge, thanks!

coveralls · 2016-06-30T09:31:43Z

Coverage decreased (-0.1%) to 99.891% when pulling 4624c0f on qt-api-ini into cef8b39 on master.

coveralls · 2016-06-30T14:40:28Z

Coverage decreased (-0.1%) to 99.891% when pulling 7f70537 on qt-api-ini into cef8b39 on master.

coveralls · 2016-06-30T20:00:24Z

Coverage decreased (-0.1%) to 99.89% when pulling 2205d78 on qt-api-ini into cef8b39 on master.

The-Compiler · 2016-06-30T20:21:47Z

heh, I guess that works too 😆

nicoddemus · 2016-06-30T20:41:07Z

I really don't have a clue why those lines are reported as missing... 😭

The-Compiler · 2016-06-30T20:53:54Z

Are they covered when you do something like PYTEST_QT_API=pyqt4v2 coverage run --source=pytestqt py.test ; coverage html locally?

I'll take another look tomorrow.

nicoddemus · 2016-07-07T13:23:22Z

@The-Compiler, I plan to merge this on the weekend after I figure out why coverage decreased.

Just wanted to know if you will find the time to get waitCallback (is that what we decided to call it?) into 2.0 as well? Otherwise after this PR is merged we can make the release. 😁

The-Compiler · 2016-07-07T18:18:56Z

Probably not, I'd expect to work on it in a few days to weeks - so let's release 2.0 without it and then do a 2.1 at some point 😉

nicoddemus · 2016-07-07T18:20:12Z

Roger! 👍

coveralls · 2016-07-21T02:13:12Z

Coverage remained the same at 100.0% when pulling 1225cdf on qt-api-ini into cef8b39 on master.

coveralls · 2016-07-21T02:20:03Z

Coverage remained the same at 100.0% when pulling d9c3d8f on qt-api-ini into 40c3e63 on master.

nicoddemus · 2016-07-21T02:24:50Z

Duh, I was looking at the wrong line.

Will merge after #141 to apply the proper changes to that branch myself.

coveralls · 2016-07-28T22:15:09Z

Coverage increased (+0.01%) to 99.897% when pulling e160714 on qt-api-ini into bad3a3b on master.

coveralls · 2016-07-28T22:40:27Z

Coverage increased (+0.01%) to 99.897% when pulling 3f7dd89 on qt-api-ini into dfb32bc on master.

coveralls · 2016-07-28T22:55:34Z

Coverage increased (+0.1%) to 100.0% when pulling 667feb2 on qt-api-ini into dfb32bc on master.

estan · 2016-08-12T12:41:11Z

@nicoddemus @The-Compiler: Thanks! The new option works great. I was the one who initially bugged @The-Compiler on IRC about this :)

nicoddemus · 2016-08-12T12:55:17Z

👍

nicoddemus mentioned this pull request May 12, 2016

Configuring Qt API in pytest.ini #129

Closed

The-Compiler modified the milestone: 2.0 Jun 25, 2016

nicoddemus force-pushed the qt-api-ini branch 2 times, most recently from 2c63ffa to 020b1b5 Compare June 29, 2016 23:50

nicoddemus changed the title ~~WIP: lazy loading qt api~~ Configure Qt API using config file Jun 29, 2016

nicoddemus reviewed Jun 29, 2016
View reviewed changes

nicoddemus force-pushed the qt-api-ini branch from 020b1b5 to 1cb63ee Compare June 30, 2016 00:04

nicoddemus force-pushed the qt-api-ini branch from 1cb63ee to 49cf8e6 Compare June 30, 2016 00:18

nicoddemus force-pushed the qt-api-ini branch 2 times, most recently from 9547d19 to a820282 Compare June 30, 2016 02:27

Qt API lazy load in qt_compat and qt_api config option

aaada5b

Fix #129

nicoddemus force-pushed the qt-api-ini branch from a820282 to aaada5b Compare June 30, 2016 08:54

The-Compiler reviewed Jun 30, 2016
View reviewed changes

nicoddemus added 2 commits June 30, 2016 06:27

Moved warning about PYTEST_QT_FORCE_PYQT to proper location in CHANGELOG

88b401f

Small fix in docstring

75cdc06

Improve docstring

7f70537

nicoddemus force-pushed the qt-api-ini branch from a0c31ff to 7f70537 Compare June 30, 2016 14:36

Disable coverage for pyqt4v2

2205d78

The-Compiler mentioned this pull request Jul 5, 2016

Runtime error when running pytest with PyQt5 application pytest-dev/pytest#1701

Closed

PyQt5 does not need to extract from QVariant values

1225cdf

Merge remote-tracking branch 'origin/master' into qt-api-ini

d9c3d8f

nicoddemus mentioned this pull request Jul 28, 2016

waitSignal and waitSignals() evaluate signal parameters #141

Merged

Merge remote-tracking branch 'origin/master' into qt-api-ini

e160714

Merge branch 'master' into qt-api-ini

3f7dd89

Coverage: validate order parameter in waitSignals

667feb2

nicoddemus merged commit 9b1a475 into master Jul 28, 2016

nicoddemus deleted the qt-api-ini branch July 28, 2016 23:14

The-Compiler mentioned this pull request Feb 6, 2017

AttributeError: '_QtApi' object has no attribute 'QtCore' #166

Closed

The-Compiler mentioned this pull request Sep 15, 2017

Integration with qtpy #184

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Configure Qt API using config file #130

Configure Qt API using config file #130

nicoddemus commented May 12, 2016 •

edited

Loading

The-Compiler commented May 13, 2016

nicoddemus commented May 13, 2016

The-Compiler commented May 13, 2016

qt_compat

nicoddemus commented May 14, 2016

ColinDuquesnoy commented May 14, 2016

nicoddemus Jun 29, 2016

coveralls commented Jun 30, 2016 •

edited

Loading

coveralls commented Jun 30, 2016

coveralls commented Jun 30, 2016 •

edited

Loading

coveralls commented Jun 30, 2016 •

edited

Loading

nicoddemus commented Jun 30, 2016

The-Compiler Jun 30, 2016

nicoddemus Jun 30, 2016

coveralls commented Jun 30, 2016 •

edited

Loading

coveralls commented Jun 30, 2016 •

edited

Loading

coveralls commented Jun 30, 2016 •

edited

Loading

The-Compiler commented Jun 30, 2016

nicoddemus commented Jun 30, 2016

The-Compiler commented Jun 30, 2016

nicoddemus commented Jul 7, 2016

The-Compiler commented Jul 7, 2016

nicoddemus commented Jul 7, 2016

coveralls commented Jul 21, 2016 •

edited

Loading

coveralls commented Jul 21, 2016 •

edited

Loading

nicoddemus commented Jul 21, 2016

coveralls commented Jul 28, 2016 •

edited

Loading

coveralls commented Jul 28, 2016 •

edited

Loading

coveralls commented Jul 28, 2016 •

edited

Loading

estan commented Aug 12, 2016

nicoddemus commented Aug 12, 2016



		@pytest.mark.parametrize('option_api', ['pyqt4', 'pyqt4v2', 'pyqt5', 'pyside'])
		def test_qt_api_ini_config(testdir, option_api):

Configure Qt API using config file #130

Configure Qt API using config file #130

Conversation

nicoddemus commented May 12, 2016 • edited Loading

The-Compiler commented May 13, 2016

nicoddemus commented May 13, 2016

qt_compat

The-Compiler commented May 13, 2016

qt_compat

nicoddemus commented May 14, 2016

ColinDuquesnoy commented May 14, 2016

nicoddemus Jun 29, 2016

Choose a reason for hiding this comment

coveralls commented Jun 30, 2016 • edited Loading

coveralls commented Jun 30, 2016

coveralls commented Jun 30, 2016 • edited Loading

coveralls commented Jun 30, 2016 • edited Loading

nicoddemus commented Jun 30, 2016

The-Compiler Jun 30, 2016

Choose a reason for hiding this comment

nicoddemus Jun 30, 2016

Choose a reason for hiding this comment

coveralls commented Jun 30, 2016 • edited Loading

coveralls commented Jun 30, 2016 • edited Loading

coveralls commented Jun 30, 2016 • edited Loading

The-Compiler commented Jun 30, 2016

nicoddemus commented Jun 30, 2016

The-Compiler commented Jun 30, 2016

nicoddemus commented Jul 7, 2016

The-Compiler commented Jul 7, 2016

nicoddemus commented Jul 7, 2016

coveralls commented Jul 21, 2016 • edited Loading

coveralls commented Jul 21, 2016 • edited Loading

nicoddemus commented Jul 21, 2016

coveralls commented Jul 28, 2016 • edited Loading

coveralls commented Jul 28, 2016 • edited Loading

coveralls commented Jul 28, 2016 • edited Loading

estan commented Aug 12, 2016

nicoddemus commented Aug 12, 2016

nicoddemus commented May 12, 2016 •

edited

Loading

coveralls commented Jun 30, 2016 •

edited

Loading

coveralls commented Jun 30, 2016 •

edited

Loading

coveralls commented Jun 30, 2016 •

edited

Loading

coveralls commented Jun 30, 2016 •

edited

Loading

coveralls commented Jun 30, 2016 •

edited

Loading

coveralls commented Jun 30, 2016 •

edited

Loading

coveralls commented Jul 21, 2016 •

edited

Loading

coveralls commented Jul 21, 2016 •

edited

Loading

coveralls commented Jul 28, 2016 •

edited

Loading

coveralls commented Jul 28, 2016 •

edited

Loading

coveralls commented Jul 28, 2016 •

edited

Loading