[css-shapes-2] Specify the grammar for shape()
#11207
Conversation
- <<position>> can be used instead of an absolute point. - One of a <<position>>'s dimension can be used for hline/vline - Since keywords define the different components, order of components is flexible. Curves: - `with` is used for control points - Two control points are separated by / - Relative control points can be relative to segment-start/segment-end/reference-box Resolution: w3c#10649 (comment) Closes w3c#10649 Closes w3c#10666
| are relative to the top-left corner of the [=reference box=], | ||
| while ''by'' indicates that the <<coordinate-pair>>s | ||
| while <css>by</css> indicates that the <<coordinate-pair>>s | ||
| are relative to the command's starting point. |
There was a problem hiding this comment.
This is no longer true when from start / from end are used. There probably needs to be a new paragraph that explains how control points are resolved, or a "see below for details".
There was a problem hiding this comment.
Right... There is a paragraph for control points, I'll refer to that.
There was a problem hiding this comment.
I think you can use ''shape()/by'' for the markup. I think.
Co-authored-by: fantasai <[email protected]>
Co-authored-by: fantasai <[email protected]>
Co-authored-by: fantasai <[email protected]>
Co-authored-by: fantasai <[email protected]>
| && <<arc-sweep>>? && <<arc-size>>? && [rotate <<angle>>]? | ||
|
|
||
| <<command-end-point>> = [ to <<position>> | by <<coordinate-pair>> ] | ||
| <<control-point>> = [ <<position>> | <<relative-control-point>> ] |
There was a problem hiding this comment.
When implementing this, I mentioned this to @smfr, and he didn't quite agree, but I thought I would bring it to a slightly larger audience.
While generally I don't really see the benefit in disallowing to be used for by (since is really just sugar for [ ], but if the distinction is going to be made, I think the grammar can be a bit more friendly:
<control-point> = [ <position> | <relative-control-point> ]
with one like so:
<absolute-control-point> = <position> [ from [ start | end | origin ] ]?
or, if you dislike the duplication, of the from section,
<control-point-from> = [ from [ start | end | origin ] ]
<absolute-control-point> = <position> <control-point-from>?
<relative-control-point> = <coordinate-pair> <control-point-from>?
This allows a user who is already using an absolute control point, for example like so:
... curve to center center with top left ...
who decides that with top left (aka with 0% 0%) should really be with top left from start to do so. In the current grammar, they have to translate their top left to percentage form and write out with 0% 0% from start.
(doing this also removes a slightly annoying, though manageable, ambiguity in the grammar when parsing <<control-point> since its not clear if "10% 10%" should match <position> or <coordinate-pair>.
There was a problem hiding this comment.
When implementing this, I mentioned this to @smfr, and he didn't quite agree, but I thought I would bring it to a slightly larger audience.
While generally I don't really see the benefit in disallowing to be used for
by(since is really just sugar for [ ], but if the distinction is going to be made, I think the grammar can be a bit more friendly:<control-point> = [ <position> | <relative-control-point> ]with one like so:
<absolute-control-point> = <position> [ from [ start | end | origin ] ]?or, if you dislike the duplication, of the
fromsection,<control-point-from> = [ from [ start | end | origin ] ] <absolute-control-point> = <position> <control-point-from>? <relative-control-point> = <coordinate-pair> <control-point-from>?
This allows to bottom left from end, how would you interpret that?
There was a problem hiding this comment.
As identical to ‘to 100% 0% from end’?
Ah so like transpose the reference rect to the start/end point... not sure it's a useful box for positioning, might create more confusion than usefulness.
There was a problem hiding this comment.
More confusing than the fact that ‘to 100% 0% from end’ is valid and ‘to bottom left’ is valid but not the merged form?
Mostly the point I am trying to make is that treating as something other than syntax sugar for two s seems undesirable. But changing this later is also non-breaking as far I can tell, so I really don’t want to push this any further than I have.
There was a problem hiding this comment.
More confusing than the fact that ‘to 100% 0% from end’ is valid and ‘to bottom left’ is valid but not the merged form?
to 0 100% from end is not valid, only with 0 100% from end or with 10px 220px / 100% 0% from end, the from start|end|origin syntax is only for curve control points.
I read with bottom left as a shorthand for with 0 100% from origin, so expanding it to with 0 100% from origin from end is indeed confusing.
Mostly the point I am trying to make is that treating as something other than syntax sugar for two s seems undesirable. But changing this later is also non-breaking as far I can tell, so I really don’t want to push this any further than I have.
Sounds good!
grammar changes:
<<position>>can be used instead of an absolute point<<position>>'s dimension can be used forhline/vlinewithis used for curve control pointsResolution: #10649 (comment)
Closes #10649
Closes #10666
[css-spec-shortname-1] Brief description which should also include the #issuenum-or-URL and/or link to relevant CSSWG minutes.
Copy the above line into the Title and replace with the relevant details. Fill in any additional details here. See https://github.com/w3c/csswg-drafts/blob/master/CONTRIBUTING.md for more info.