| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
|
|
Upgrade PyVirtualDisplay
Update virtualenv from 20.0.21 to 20.0.24
Update pytest-xvfb from 1.2.0 to 2.0.0
Update pytest-instafail from 0.4.1.post0 to 0.4.2
Update pytest-cov from 2.9.0 to 2.10.0
Update py-cpuinfo from 5.0.0 to 6.0.0
Update py from 1.8.1 to 1.8.2
Update more-itertools from 8.3.0 to 8.4.0
Update hypothesis from 5.16.0 to 5.18.0
Update sphinx from 3.1.0 to 3.1.1
Update setuptools from 47.1.1 to 47.3.1
Update mypy from 0.780 to 0.781
Update pep8-naming from 0.10.0 to 0.11.1
Update flake8 from 3.8.2 to 3.8.3
Update requests from 2.23.0 to 2.24.0
Update codecov from 2.1.4 to 2.1.7
Update certifi from 2020.4.5.2 to 2020.6.20
|
|
Thanks to Ingvix in IRC for the ASCII-art! :)
|
|
|
|
|
|
|
|
Recreate the inspector when it was shown inline and gets detached to work
around an issue where QtWebEngine (for some odd reason) doesn't show a window
at all.
See #5540
|
|
|
|
|
|
|
|
This reverts commit 69a225bcf2a62207bb13d6c2d7c9cd1726146909.
Fails tests and ends up being the more complex solution.
|
|
|
|
It doesn't work properly: If we detach, we'd also need to re-attach when
reopening the inspector!
Let's just not do it for now and see what people think. That also has some
benefits like the inspector remembering what tab the user was on when closing
it.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This way we don't need to have if's everywhere, and the two classes only have
some 3 lines in common anyways!
|
|
Closes #5501
|
|
The general consensus seems to be that the inspector is the first tab of the
devtools, and Chromium also calls them devtools everywhere.
Closes #5438
|
|
|
|
|
|
|
|
This way we don't have to care about things like sizeHint manually.
|
|
Don't only reparent the widget, also make sure we don't actually control (and
e.g. resize) it anymore.
|
|
Qt 5.8 support was removed in April 2018 with
69a58c959705b2dd469ba416250f884472593bee, and the last crash report I got with
Qt 5.8 was in the same month.
The workaround removed in this commit still were around, and despite of what
the version check says, they are *not* needed with Qt 5.7.
Those were workarounds for the following bugs:
- https://bugreports.qt.io/browse/QTBUG-58698
(#2290, 1af951fd62b4878b2eaeb63e628b0ce5c06eeae8)
- https://bugreports.qt.io/browse/QTBUG-58982
(#2261, 90f472bf590b6fdb3e7f4158ae420002ccbdee17)
See #3608
|
|
This was true for QtWebKit and doesn't apply for QtWebEngine anymore.
See #730
|
|
|
|
|
|
|
|
Fixes #4348
|
|
|
|
|
|
|
|
|
|
|
|
This reverts commit 64ffce27f9348e14836528c0313d3048669a29c7.
|
|
|
|
The inspector object is now re-used and only created once (when the inspector
is opened first). This also moves some logic from CommandDispatcher to the
inspector object.
The geometry is now saved again when the window is closed (via closeEvent)
rather than in AbstractWebInspector.set_position - after all, the window will
also close when the position is changed.
Finally, set_position(None) changed its meaning: Its now not used anymore to
close/delete the inspector, but instead means "use the last saved position".
|
|
- Use a smaller size (600 instead of 1000) as base so that the test works with
smaller screens (e.g. with macOS on Travis, the screen resolution is
1024x768, with 658px being available vertically).
- Skip the test instead of failing when resizing doesn't work (e.g. when
running under a tiling WM without Xvfb)
- Fix handle calculation if the handle has an odd number of pixels (e.g. 5px on
macOS).
|
|
|
|
This is a follow-up to f0eed465ab608061212366c6ebc26a0c47803060
See #4657
|
|
See #2377
|
|
|
|
Closes #5451
|
