Re: [whatwg] Splitting CanvasRenderingContext2D

[Ian Kilpatrick]

Great, thanks all.

I'll put together the start of a PR this week (just modifying the IDL) so
that we can start discussing name / where everything goes.

Thanks,
Ian

On Tue, Dec 1, 2015 at 12:11 PM, Justin Novosad  wrote:

> I think it may be possible to do this refactor without changing the order
> of things since the current IDL has things grouped into categories that
> more or less map to what Ian is suggesting.
>
> I am currently in the middle of working on a pull request for the
> OffscreenCanvas propsal, and I have to say I would very much like to rebase
> it on top of what Ian is suggesting so that I could easily define an
> "OffscreenRenderingContext2D" that excludes or redefines APIs that interact
> with the Document or view.
>
> On Tue, Dec 1, 2015 at 11:55 AM, Boris Zbarsky  wrote:
>
> > On 12/1/15 11:40 AM, Domenic Denicola wrote:
> >
> >> Since it has no web-visible impact
> >>
> >
> > That may or may not be true.  For example, it's quite likely to change,
> in
> > practice, the enumeration order for properties on the obect.  Of course
> > there is no spec defining that enumeration order right now... but in
> > practice I believe UAs mostly do it in the IDL order.
> >
> > Normally I wouldn't worry about it, but there is known content out there
> > that depends on the enumeration order of _some_ things on this particular
> > interface.  See https://bugzilla.mozilla.org/show_bug.cgi?id=631117 and
> > https://bugzilla.mozilla.org/show_bug.cgi?id=623437 (note that these are
> > about different URLs, though caused by the same underlying issue).
> >
> > In practice I expect that an interface split that keeps arc() and arcTo()
> > together in this case will be OK.  I just want to point out that the risk
> > here is sadly nonzero.
> >
> > -Boris
> >
>

Re: [whatwg] Splitting CanvasRenderingContext2D

[Boris Zbarsky]

On 12/1/15 11:40 AM, Domenic Denicola wrote:

Since it has no web-visible impact


That may or may not be true.  For example, it's quite likely to change, 
in practice, the enumeration order for properties on the obect.  Of 
course there is no spec defining that enumeration order right now... but 
in practice I believe UAs mostly do it in the IDL order.


Normally I wouldn't worry about it, but there is known content out there 
that depends on the enumeration order of _some_ things on this 
particular interface.  See 
https://bugzilla.mozilla.org/show_bug.cgi?id=631117 and 
https://bugzilla.mozilla.org/show_bug.cgi?id=623437 (note that these are 
about different URLs, though caused by the same underlying issue).


In practice I expect that an interface split that keeps arc() and 
arcTo() together in this case will be OK.  I just want to point out that 
the risk here is sadly nonzero.


-Boris

Re: [whatwg] Splitting CanvasRenderingContext2D

[Domenic Denicola]

From: whatwg [mailto:whatwg-boun...@lists.whatwg.org] On Behalf Of Ian 
Kilpatrick

> There are a couple of ways to split up the CanvasRenderingContext2D API,
> namely adding inheritance + [NoInterfaceObject] to superclasses (as to not
> change prototype chain, maybe not a large web compat concern), or to
> create additional interfaces and get *RenderingContext2D to implement.

Split the interfaces seems fine. Since it has no web-visible impact and is just 
a spec refactoring I anticipate any such change will be quickly LGTMed. In 
general, we're greatly in favor of any integration points like this we can add 
to HTML that simply consist of spec refactorings to allow better integration 
with other specs.

> Thoughts on this as a PR to the HTML spec?
> If so any preferences for how this is done?

https://github.com/whatwg/html#pull-requests should help get you started!