Clipping for OffsetBoxes #4466

has2k1 · 2015-05-25T02:40:05Z

Child Artists of DrawingArea now get clipped to the bounds of the parent

For reviewers

Also related to Offsetbox is another commit (or PR whichever is best) that will follow. This will include Offsetboxes in the tight_layout. In the tests it would make sense to cover invisible offsetboxes anchored to (in)visible axes. These tests would probably be sufficient for PR tight_layout ignores invisible axes #2008. Is O.K. to cherry-pick that PR?

has2k1 · 2015-05-25T03:18:04Z

An example

examples/pylab_examples/anchored_artists.py
66:        self.da = DrawingArea(width, height, xdescent, ydescent, clip=True)

used an API term (clip=True) that was not explained in the docs and not implemented in the source. I got rid of the keyword (i.e an API change) because making it part of the implementation was not convincing. Would it mean, clip the DrawingArea (I could not think of a use case) or clip the children (which would have their own clip status).

Would this situation better be suited for a deprecation warning?

tacaswell · 2015-05-25T04:24:21Z

lib/matplotlib/offsetbox.py

-        # Clipping has not been implemented in the OffesetBox family, so
-        # disable the clip flag for consistency. It can always be turned back
-        # on to zero effect.
-        self.set_clip_on(False)


probably should leave this.

tacaswell · 2015-05-25T04:36:51Z

I think that a cilp on the offset box does make sense, the path of the DrawArea would first be clipped to it's clip path and then the artist would be clipped to that path. The use case I am thinking of is an OffsetBox which is half outside of an Axes, but I see the case that that should be a different PR.

This looks reasonable to me, but I would prefer to keep the clip kwarg to turn on/off the clipping of the children to the draw area and setting it to False by default. I think the change to clipping to the draw area by default is a much bigger API change than dropping the clip input.

tacaswell · 2015-05-25T04:37:18Z

doc/api/api_changes/2015-04-18-drawingarea.rst

+was done with the `clip` argument. The object did not do any clipping
+regardless of that parameter. Now the object can and does clip the child `Artists` if they are set to be clipped.
+
+You can turn off the clipping using :method:`Artist.set_clip_on(False)`


Can you be more explicit about which artists need to be changed?

No Artist needs to be changed. Maybe I will use a more explicit example.

By changed, I mean what artist do you call set_clip_on(False) on. It is not clear from this text that it should be on the children, not on the DrawingArea.

has2k1 · 2015-05-25T05:13:15Z

On the clipping expectation, I had gone for consistency with children added to the Axes vs maintaining past code behaviour, provided that the tests passed.

tacaswell · 2015-05-26T00:52:17Z

doc/api/api_changes/2015-04-18-drawingarea.rst

@@ -0,0 +1,8 @@
+'OffsetBox.DrawingArea' no longer accepts the 'clip' keyword argument


The title is no longer accurate. Should be 'OffsetBox.DrawingArea now respects clip' or something to that effect.

tacaswell · 2015-05-26T00:55:28Z

That is a fair point and if we were adding these classes from scratch I think I would agree with your original implementation, but this code has been in the wild for a long time and maintaining current behavior is a very high priority.

Other than a few comments on the documentation, this looks good!

- Child `Artists` of `DrawingArea` can now get clipped to the bounds of the parent

has2k1 · 2015-05-26T03:11:28Z

It is fixed up.

Plus, are there any objections to Offsetboxes being considered in the tight_layout?

tacaswell · 2015-05-26T03:13:48Z

I have no protests to including OffsetBoxes in tight layout considerations. My only concern is sinking too much time into the current tight_layout code when it might be better directed to sorting out a layout manager based on a linear constraint solver.

tacaswell · 2015-05-26T03:16:08Z

lib/matplotlib/offsetbox.py

@@ -569,14 +570,16 @@ class DrawingArea(OffsetBox):
    """
    The DrawingArea can contain any Artist as a child. The DrawingArea
    has a fixed width and height. The position of children relative to
-    the parent is fixed.
+    the parent is fixed. The children can be  clipped at the


super picky, extra space between 'be' and 'clipped'

I will take care of this. I want to extend this PR a bit and will just do it my self rather than explain what I want.

Well, I rebased and pushed already.

Sorry, it looks like I won the race condition on that and merged the previous version.

And if you have a change set, please open a PR.

On Mon, May 25, 2015 at 11:25 PM Hassan Kibirige notifications@github.com
wrote:

In lib/matplotlib/offsetbox.py
#4466 (comment):

@@ -569,14 +570,16 @@ class DrawingArea(OffsetBox):
"""
The DrawingArea can contain any Artist as a child. The DrawingArea
has a fixed width and height. The position of children relative to

the parent is fixed.

the parent is fixed. The children can be clipped at the

Well, I rebased and pushed already.

—
Reply to this email directly or view it on GitHub
https://github.com/matplotlib/matplotlib/pull/4466/files#r31002511.

ENH/API: Clipping for OffsetBoxes

tacaswell · 2015-05-26T03:23:40Z

Thanks!

has2k1 · 2015-05-26T03:34:10Z

I already have a changeset for the tight_layout. On the linear constraint solver, has this been discussed elsewhere?

tacaswell · 2015-05-26T03:44:33Z

The discussion is a bit scattered, most of it is at #1109

tacaswell added the status: needs review label May 25, 2015

has2k1 force-pushed the fix-offsetbox-clipping branch from e31da20 to e477723 Compare May 25, 2015 02:59

has2k1 force-pushed the fix-offsetbox-clipping branch 2 times, most recently from ff44733 to 2c54249 Compare May 25, 2015 04:00

tacaswell added this to the next point release milestone May 25, 2015

tacaswell reviewed May 25, 2015
View reviewed changes

has2k1 force-pushed the fix-offsetbox-clipping branch from 2c54249 to a9d9084 Compare May 25, 2015 05:48

tacaswell reviewed May 26, 2015
View reviewed changes

has2k1 force-pushed the fix-offsetbox-clipping branch 2 times, most recently from afef2b2 to 6ade0dd Compare May 26, 2015 01:35

Clipping for OffsetBoxes

ea082aa

- Child `Artists` of `DrawingArea` can now get clipped to the bounds of the parent

has2k1 force-pushed the fix-offsetbox-clipping branch from 6ade0dd to ea082aa Compare May 26, 2015 02:36

tacaswell reviewed May 26, 2015
View reviewed changes

tacaswell added a commit that referenced this pull request May 26, 2015

Merge pull request #4466 from has2k1/fix-offsetbox-clipping

420f7c9

ENH/API: Clipping for OffsetBoxes

tacaswell merged commit 420f7c9 into matplotlib:master May 26, 2015

tacaswell removed the status: needs review label May 26, 2015

tacaswell mentioned this pull request Jun 13, 2015

ENH: property if DrawingArea clips children #4473

Merged

has2k1 mentioned this pull request Feb 17, 2023

Save file with actual pixel dimensions has2k1/plotnine#662

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Clipping for OffsetBoxes #4466

Clipping for OffsetBoxes #4466

has2k1 commented May 25, 2015

has2k1 commented May 25, 2015

tacaswell May 25, 2015

tacaswell commented May 25, 2015

tacaswell May 25, 2015

has2k1 May 25, 2015

tacaswell May 25, 2015

has2k1 commented May 25, 2015

tacaswell May 26, 2015

tacaswell commented May 26, 2015

has2k1 commented May 26, 2015

tacaswell commented May 26, 2015

tacaswell May 26, 2015

tacaswell May 26, 2015

has2k1 May 26, 2015

tacaswell May 26, 2015

tacaswell May 26, 2015

tacaswell commented May 26, 2015

has2k1 commented May 26, 2015

tacaswell commented May 26, 2015

		@@ -0,0 +1,8 @@
		'OffsetBox.DrawingArea' no longer accepts the 'clip' keyword argument

Clipping for OffsetBoxes #4466

Clipping for OffsetBoxes #4466

Conversation

has2k1 commented May 25, 2015

has2k1 commented May 25, 2015

Choose a reason for hiding this comment

tacaswell commented May 25, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

has2k1 commented May 25, 2015

Choose a reason for hiding this comment

tacaswell commented May 26, 2015

has2k1 commented May 26, 2015

tacaswell commented May 26, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

tacaswell commented May 26, 2015

has2k1 commented May 26, 2015

tacaswell commented May 26, 2015