Re: [css-font-loading] controlling font loading display

Ilya Grigorik wrote:

> John Daggett wrote:
>
> > I'd like to propose a simpler property for controlling font display
> > while downloadable font resources are loading. I think the previous
> > proposal [1] is overly complex and the value names aren't terribly
> > intuitive.
>
> John, can you elaborate on which parts are overly complex, and why? Is
> it the timers?

Yes, having explicit timeout values seems overly specific. The effect of
timeouts will vary across implementations based on device capabilty,
load prioritization and reflow strategy so it's hard to imagine authors
actually getting consistent results with specific timeout values. Better
I think to allow authors to define the overall load strategy and let the
browser handle the details.

> the font rendering strategy

Sorry to nitpick but I think we really need to avoid using this
terminology. The property under discussion here is a way to control text
display when font resources aren't available, it has no effect on how
fonts are actually rendered on the screen (i.e. it has no effect on
antialiasing, hinting, subpixel spacing, etc.).

> One thing I'm wondering about is whether we can/should uncouple timers
> from rendering strategy:
> - the font rendering strategy varies based on particular content on
>   the page; it's a property on the content, not the font.
> - the font fetch timeout is a property on the font resource.

As I mentioned above, I'm not really keen on specific timeout values. I
think it also might make sense to define the 'blank-fallback' timeout to
use a timeout relative to the page, for example navigationStart, instead
of making it a function of when the initial load started. That way the
time to when the user sees fallback text is more consistent across pages.

Cheers,

John Daggett

​