[whatwg] Adding descent to TextMetrics

[Charles Pritchard]

On Aug 10, 2010, at 4:09 PM, Ian Hickson <ian at hixie.ch> wrote:

> On Thu, 8 Jul 2010, Charles Pritchard wrote:
>> 
>> We've been doing some work with html line boxes; mainly, mixed font 
>> styles in a text area, using the HTML Canvas element. While Canvas 
>> fillText does have the textBaseline attribute, in order to set the 
>> current alignment, it does not provide enough metrics through the 
>> measureText function for us to exploit the textBaseline attribute across 
>> fonts (sizes and styles).
>> 
>> Adding descent, to the measureText object would allow us to do more. 
>> Currently, TextMetrics only returns "width". I'm proposing adding 
>> "descent" to that list.
> 
> I recommend not using <canvas> for text editing.

I've worked on a substantial amount of code dealing with text editing. At present, the descent of the current font has been the only deficiency.

I feel that using Canvas to implement HTML5/CSS provides a quality proof of the completeness of the 2D API.

> 
> Adding more features to TextMetrics is on the cards for a future version 
> of the canvas API, but at the moment I'm waiting until more of the spec is 
> reliably implemented before adding more features, so that we don't get too 
> far ahead of the spec.


We're fine simply including a table of descent metrics (they're essentially a single static value for each font).

Exposing the metric value is a very simple process for implementors (low cost in time to implementation).

I understand that the Canvas spec is essentially frozen (though this seems like a deficiency in the standard [why have textBaseline without it?]) -- I'd like to see some place to discuss the  future standard, so as to avoid x-vend-descent as the property value.



> 
> -- 
> Ian Hickson               U+1047E                )\._.,--....,'``.    fL
> http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
> Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'