Re: [css-fonts] Proposal for 'font-rendering' property

On Thu, Apr 9, 2015 at 4:16 PM, Tab Atkins Jr. <jackalmage@gmail.com> wrote:

> I think this might only be accessible to @google.com addresses.  If
> so, I'll bug Ilya to loosen it up. The gist is that our analytics
> shows that ~30% of page loads show at least some "blank text" *in the
> viewport* (that is, text that actually matters, that impacts the
> user's experience).
>
> Plus, given a conservative estimate of a 650ms budget from
> first-byte-received to page-is-ready-to-display-something (based on
> actual analysis), even a small font (10kb) will often take longer than
> the budget to download.  This means that even if we started the font
> download upon receiving the first byte of the response, we'll still
> often get blank text for some period of time.
>

Public doc/post with aggregated numbers for Chrome for Android:
https://www.igvita.com/2015/04/10/fixing-the-blank-text-problem/

+1 to Tab's comments. Optimizing filesize and initiating font downloads
sooner are important but not sufficient to resolve the larger problem. We
need to give developers control over which rendering strategy is applied to
different content on the page.

ig