Re: FW: [Bug 27638] New: FEATURE REQUEST: Selectively enable metrics in measureText using optional second argument

Interesting idea. I would suggest using a dictionary rather than an
interface for CanvasMeasureTextOptions.

Slightly OT:
When reviewing code related to the measureText implementation, something
that struck me is that not all fonts provide all of the information for the
height metrics, so sometimes the implementation is forced to make educated
guesses based on the metrics that are available in the font file (e.g
applying ratios to metrics that are provided). When you say that your
polyfill is robust, are you implying that you have a solid solution to
filling in the missing metrics?  I think this particular problem is an area
that is at risk of being non-interoperable between browsers, and perhaps
should be spec'ed.  What do you think?

On Wed, Dec 17, 2014 at 10:40 AM, Paul Cotton <Paul.Cotton@microsoft.com>
wrote:

> FYI.
>
> Paul Cotton, Microsoft Canada
> 17 Eleanor Drive, Ottawa, Ontario K2E 6A3
> Tel: (425) 705-9596 Fax: (425) 936-7329
>
>
> -----Original Message-----
> From: bugzilla@jessica.w3.org [mailto:bugzilla@jessica.w3.org]
> Sent: Wednesday, December 17, 2014 7:35 AM
> To: public-html-admin@w3.org
> Subject: [Bug 27638] New: FEATURE REQUEST: Selectively enable metrics in
> measureText using optional second argument
>
> https://www.w3.org/Bugs/Public/show_bug.cgi?id=27638
>
>             Bug ID: 27638
>            Summary: FEATURE REQUEST: Selectively enable metrics in
>                     measureText using optional second argument
>            Product: HTML WG
>            Version: unspecified
>           Hardware: All
>                 OS: All
>             Status: NEW
>           Severity: enhancement
>           Priority: P2
>          Component: HTML Canvas 2D Context
>           Assignee: jaymunro@microsoft.com
>           Reporter: motiz88@gmail.com
>         QA Contact: public-html-bugzilla@w3.org
>                 CC: mike@w3.org, public-html-admin@w3.org,
>                     public-html-wg-issue-tracking@w3.org
>
> Hello,
>
> I am the author of https://github.com/motiz88/canvas-text-metrics-polyfill
> .
>
> I'd like to make the case for a small enhancement following up on the
> resolution of https://www.w3.org/Bugs/Public/show_bug.cgi?id=7798 ,
> relating to the properties added to TextMetrics (
> https://html.spec.whatwg.org/#drawing-text-to-the-bitmap ).
>
> Background: Browser support for this useful and simple API is still
> surprisingly scarce (to my knowledge only Chrome Canary has it, behind an
> experimental flag, with a buggy implementation), so I have written a
> polyfill (which I believe is the first conforming to this API) based mainly
> on off-screen drawing and scanning. This approach is rather slow in the
> general case, but results in a robust and (at least, largely) conforming
> polyfill.
>
> My suggested enhancement below, already implemented in my polyfill, would
> enable such polyfills (and possibly native implementations?) to be lighter
> and faster, in cases where e.g. only one or two metrics are needed out of
> the whole set. The availability of a fully conforming and performant
> polyfill may finally spur adoption of this API by developers and browser
> makers.
>
> <suggestion>
>
> interface CanvasRenderingContext2D {
> ...
> TextMetrics measureText(DOMString text, optional CanvasMeasureTextOptions
> options); ...
> };
>
> interface CanvasMeasureTextOptions {
>   // x-direction
>   attribute boolean width; // advance width
>   attribute boolean actualBoundingBoxLeft;
>   attribute boolean actualBoundingBoxRight;
>
>   // y-direction
>   attribute boolean fontBoundingBoxAscent;
>   attribute boolean fontBoundingBoxDescent;
>   attribute boolean actualBoundingBoxAscent;
>   attribute boolean actualBoundingBoxDescent;
>   attribute boolean emHeightAscent;
>   attribute boolean emHeightDescent;
>   attribute boolean hangingBaseline;
>   attribute boolean alphabeticBaseline;
>   attribute boolean ideographicBaseline; };
>
> Augmented semantics of measureText():
>
> If the options argument is missing or null, measureText() behaves as
> currently specified, and must return a TextMetrics object with valid values
> set on all its specified properties.
>
> If there is a non-null options argument, then the returned TextMetrics
> object is only required to contain the set of properties whose names
> coincide with the names of properties on the options object which are both
> (1) set, and (2) equal to true. The TextMetrics object MAY additionally
> contain some properties it is not required to, in which case their values
> are not defined.
>
> </suggestion>
>
> I would appreciate any feedback on this.
>
> Thanks,
>
> Moti Zilberman
>
> --
> You are receiving this mail because:
> You are on the CC list for the bug.
>
>