added modifier key tracking in MouseEvents #6159
Conversation
|
Do you want to take a crack at making this work for the other backends? |
lib/matplotlib/backend_bases.py
Outdated
| @@ -1886,7 +1891,7 @@ def scroll_event(self, x, y, step, guiEvent=None): | |||
| step=step, guiEvent=guiEvent) | |||
| self.callbacks.process(s, mouseevent) | |||
|
|
|||
| def button_press_event(self, x, y, button, dblclick=False, guiEvent=None): | |||
| def button_press_event(self, x, y, button, dblclick=False, modifiers=None, guiEvent=None): | |||
|
I'll take a look. |
| @@ -254,8 +254,11 @@ def mousePressEvent(self, event): | |||
| # flipy so y=0 is bottom of canvas | |||
| y = self.figure.bbox.height - event.pos().y() | |||
| button = self.buttond.get(event.button()) | |||
| modifiers = "+".join([name for name, mod_key, qt_key in MODIFIER_KEYS | |||
| if (int(event.modifiers()) & mod_key) == mod_key]) | |||
There was a problem hiding this comment.
Maybe make modifiers a set instead of a string like "ctrl+alt"? Also, I think this could be simplified to
name for name, mod_key, _ in MODIFIER_KEYS if event.modifiers() & mod_key
There was a problem hiding this comment.
Since 2.1 is dropping 2.6 support, you write that in a set comprehension style.
There was a problem hiding this comment.
We have already dropped 2.6 support (but test it on the 1.5.x because we are nice).
|
You'll want to update the docs as well, both to mention the existence of [FYI for context @mdboom @tacaswell: @jerryz123's an undergrad working with @stefanv and me on viscm-stuff, and this came out of that.] |
9c93c36 to
5ee60ba
Compare
|
Could someone take a look at this? I've added support for tk, gtk, qt, and wx. |
|
@tacaswell Do you know who is familiar enough with the mac backend to give Jerry a hand? |
|
@mdboom probably. The mac backend just had some (major!) changes land. It On Mon, Apr 4, 2016 at 10:43 PM Stefan van der Walt <
|
5ee60ba to
1d4569b
Compare
|
The mac backend appears to use modifiers to alter the button clicked. A Ctrl+Click registers as a right button click, while an Alt+Click registers as a middle button click. Should this be changed? |
|
I am leaning towards yes on changing the behavior of the mac backend. The single-button external mice seem to have died and I think on all of the fancy trackpads are now multi touch (so 2 and 3 finger taps/clicks are the right/middle mouse events). |
|
@jerryz123 Are you still interested in working on this? |
6d441fa to
00ea2fb
Compare
00ea2fb to
4ea0405
Compare
|
@tacaswell Sorry for the delay. I have adjusted the mac behavior so that modifier keys are preserved on mouse clicks. What would be the correct way to get test coverage over modifier key behavior? |
|
We have some test of the qt5 backend which feed synthetic events through. We don't have good tests for the other backends other than manually testing by devs. |
|
And the delay is as much on our side! |
|
This seems useful if @jerryz123 or someone else had time to rebase and get it working again... |
|
Superseded by #23473. Thanks for initiating the work on this! |
Problem:
Currently, modifier keys in MouseEvents are only recognized if the FigureCanvas received a KeyEvent with the MouseEvent. However, in situations where the FigureCanvas does not receive keyboard focus, no KeyEvent is received, and thus the modifier-key is lost.
Solution
Modifier-key info should be carried in the MouseEvent created by a button_press_event. Add a modifiers field to mouse_events to track modifier-key status. I have updated the qt5 backend to account for this, but this can easily be extended to other backends.