Re: EOT-Lite File Format

On Thu, 2009-07-30 at 21:53 -0500, Tab Atkins Jr. wrote:
> On Thu, Jul 30, 2009 at 9:45 PM, Thomas Lord<lord@emf.net> wrote:
> > On Fri, 2009-07-31 at 02:06 +0000, Sylvain Galineau wrote:
> >> >The EOTL proposal says "is not loaded" if the
> >> >root string is non-nil.  That's a rootstring check.
> >> >It is very distinct from ignoring the rootstring,
> >> >at least as stated.
> >>
> >> The proposal is here.
> >>
> >> http://lists.w3.org/Archives/Public/www-font/2009JulSep/0780.html
> >>
> >> There is no rootstring check.
> >>
> >> The header version proposed in the latest amendment has no rootstring.
> >>
> >> Last chance.
> >
> >
> > I'm confused because on the one hand you say there is
> > no rootstring and on the other you say that EULAs might
> > require the rootstring to be set so that older versions
> > of IE enforce a simulacrum of appropriate origin restrictions.
> >
> > I would like a clear, positive statement that the intent
> > here is that a UA may come across a font file which
> > contains a non-nil root string, where that root string
> > does not match the URL of the page linking to that font, and that
> > the UA may then go ahead and render with that font anyway
> > without, in doing so, being non-conforming.  This behavior
> > of a UA should not only be permissible, but suggested ("SHOULD").
> 
> A conforming UA MUST ignore the rootstring.  Phrased probably a little
> better it MUST treat certain parts of the header (specified precisely
> by people who know the details a little better than me, informally
> being everything that isn't explicitly checked) as meaningless padding
> and MUST NOT take any action based on information from those sections
> of the header.

As I said, you are braver than I am in that regard.
My thought was SHOULD rather than MUST so that older
IE gets retroactive conformance (else what's the point
of EOT-lite at all?).



> 
> ~TJ

Received on Friday, 31 July 2009 03:02:37 UTC