1266316 - Maybe change our serialization of clip-path shapes once the spec defines what should happen

Status: RESOLVED WORKSFORME

(Core :: Layout, defect, P3)

Description

[Jonathan Watt [:jwatt]]

Once the following is addressed in the spec we should modify our serialization of clip-path shapes if appropriate.

https://lists.w3.org/Archives/Public/public-fx/2016AprJun/0000.html
https://github.com/w3c/fxtf-drafts/issues/3

Comment 1

[Jonathan Watt [:jwatt]]

Attachment: strawman patch

If the decision is that we should omit the <geometry-box> when a shape is present that causes a problem for us. We would want to ideally change this code something like this, but at this point we have no knowledge of whether a shape was specified too.

Comment 2

[Jonathan Watt [:jwatt]]

Attachment: strawman patch

Comment 3

[Astley Chen (inactive)]

Jonathan, r u still currently working on this ?

Comment 4

[Jonathan Watt [:jwatt]]

I'm not, no. Do you have someone who could take it?

Comment 5

[Astley Chen (inactive)]

Yes, will do.

Comment 6

[Astley Chen (inactive)]

jwatt, from the last feedback from dirk on GitHub, do you have any other concerns ?


Besides, I've done couple tests on <basic-shape> serialization for clip-path, it seems the result is a bit different from what spec defined[1]. If we want to align, I will create follow-up bugs to do that.

[1]examples in https://www.w3.org/TR/css-shapes-1/#basic-shape-serialization

Comment 7

[u459114]

(In reply to Astley Chen [:astley] (UTC+8) from comment #6)
> jwatt, from the last feedback from dirk on GitHub, do you have any other
> concerns ?
> 
> 
> Besides, I've done couple tests on <basic-shape> serialization for
> clip-path, it seems the result is a bit different from what spec defined[1].
> If we want to align, I will create follow-up bugs to do that.
> 
> [1]examples in https://www.w3.org/TR/css-shapes-1/#basic-shape-serialization

Could you update your test cases here?

Comment 8

[u459114]

https://drafts.csswg.org/css-shapes-1/#shape-outside-property

Computed value:
  as defined for <basic-shape> (with <shape-box> following, if supplied)...

In current implementation, clip-path's behavior is aligned with shape-outside.  Unless spec changes to what jwatt mentioned in comment 1, current implementation of reference box serialization is correct.

Comment 9

[u459114]

And I don't think this issue block clip-path basic shape shipping.

Comment 10

[Jonathan Watt [:jwatt]]

(In reply to Astley Chen [:astley] (UTC+8) from comment #6)
> jwatt, from the last feedback from dirk on GitHub, do you have any other
> concerns ?

I'll take another look at this today or tomorrow. I need to page the background to this issue back in and think about what Dirk said.

Comment 11

[Astley Chen (inactive)]

I'll suggest filing a spec issue to follow up the discussion here.

Comment 12

[Jonathan Watt [:jwatt]]

Looks okay now.