1486646 - The relativeness in the interpolation results of SVG path (d property/attribute) doesn't match those in web-platform-tests

Status: NEW

(Core :: SVG, enhancement, P5)

Description

[Boris Chiou [:boris]]

while working on Bug 1486094, I notice that the relativeness in interpolation results of svg paths is always "absolute" [1]. That's why we didn't pass all the tests in wpt because we always convert the relativeness of the start segment according to the relativeness of the end segment [2].

[1] https://searchfox.org/mozilla-central/rev/55da592d85c2baf8d8818010c41d9738c97013d2/testing/web-platform/tests/svg/path/property/d-interpolation-relative-absolute.svg#18-29 

[2] https://searchfox.org/mozilla-central/rev/55da592d85c2baf8d8818010c41d9738c97013d2/dom/svg/SVGPathSegListSMILType.cpp#317-319

Comment 1

[Boris Chiou [:boris]]

Oh. I guess this is not the only reason that we fail all the tests for svg path d property. The expectations use "path('M...')", instead of 'M...'. It seems d property only accepts a string (i.e. without "path" function keyword)...

Comment 2

[Boris Chiou [:boris]]

Downgrade the priority because the wpt seems buggy.

Comment 3

[Boris Chiou [:boris]]

Temporarily let this bug block svg2, for now. (We don't have svg1.1 meta bug, and I'm not sure if svg2 wants to support path keyword in d property. Or this is just a wpt bug. :()

Comment 4

[Boris Chiou [:boris]]

Temporarily let this bug block svg2, for now. (Please feel free to update this.)

Comment 5

[Boris Chiou [:boris]]

(In reply to Boris Chiou [:boris] from comment #1)
> Oh. I guess this is not the only reason that we fail all the tests for svg
> path d property. The expectations use "path('M...')", instead of 'M...'. It
> seems d property only accepts a string (i.e. without "path" function
> keyword)...

The discussion: https://github.com/w3c/svgwg/issues/320

Comment 6

[Boris Chiou [:boris]]

(In reply to Boris Chiou [:boris] from comment #5)
> (In reply to Boris Chiou [:boris] from comment #1)
> > Oh. I guess this is not the only reason that we fail all the tests for svg
> > path d property. The expectations use "path('M...')", instead of 'M...'. It
> > seems d property only accepts a string (i.e. without "path" function
> > keyword)...
> 
> The discussion: https://github.com/w3c/svgwg/issues/320

And it seems we should support "path(...)" for d property, according to the result of recently discussion: (https://github.com/w3c/svgwg/issues/320#issuecomment-414462645). Anyway, this should be a separate bug.

Comment 7

[Boris Chiou [:boris]]

Sorry, let me rewrite the description:

While working on Bug 1486094, I notice that the relativeness in interpolation results of svg paths is always "absolute" [1]. There is a discussion about this [2]. The current implementation of svg path in Gecko always converts the start segment according to the end segment type [3]. In Bug 1486094 I may convert the SVG Path into absolute format during interpolation, and then serialize them as absolute segments, so we could pass the current wpt tests. Probably we have to update the interpolation for SVG d attribute in this bug.

[1] https://searchfox.org/mozilla-central/rev/55da592d85c2baf8d8818010c41d9738c97013d2/testing/web-platform/tests/svg/path/property/d-interpolation-relative-absolute.svg#18-29
[2] https://github.com/w3c/svgwg/issues/321
[3] https://searchfox.org/mozilla-central/rev/55da592d85c2baf8d8818010c41d9738c97013d2/dom/svg/SVGPathSegListSMILType.cpp#317-319