Re: <canvas> JPEG quality

On Tue, 29 Jul 2008, Boris Zbarsky wrote:
>
> Ian Hickson wrote:
> > The extensibility mechanism is for the spec, not the implementors. :-)
> 
> That just sidesteps the issue, since implementors are expected to 
> implement before the spec if finalized, and hence would be likely to 
> ship implementations that they would then have to change...  The result 
> is not unlikely to be that the earlier release throws on input that the 
> later release accepts.

I have been extremely careful about not changing things that have already 
shipped, or, when changing them, making sure that the changes can co-exist 
with the previous implementation. See, e.g., the globalStorage changes, 
which were done while simultaneously changing the name of the API to 
localStorage to avoid clashes.


> > and the other is to discuss experiments on one of the HTML5 mailing 
> > lists, so that all the other vendors (and myself) are aware of what is 
> > going on, and can compare notes. Either would be fine. (The latter is 
> > what Opera is doing with image/jpeg.)
> 
> How does that help the situation with not following the letter of the 
> spec, though?

By keeping everyone in the loop, we can avoid clashes and 
incompatibilities.

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

Received on Wednesday, 30 July 2008 01:19:50 UTC