- From: Sylvain Galineau <sylvaing@microsoft.com>
- Date: Tue, 11 Nov 2008 17:11:08 -0800
- To: Thomas Lord <lord@emf.net>
- CC: Dave Crossland <dave@lab6.com>, "www-style@w3.org" <www-style@w3.org>
> From: Thomas Lord [mailto:lord@emf.net] > Hi, Sylvain. > > > On Tue, 2008-11-11 at 10:37 -0800, Sylvain Galineau wrote: > > I had not - so thank you - but I am not sure > > this answers my much more modest and general > > question. Let me retry. > > > > Beyond raw TT/OT and a compressed version of > > them, a new font format could well emerge five > > years from now. Are user agents supposed to > > sniff the content returned by the server to > > figure out which flavor the font is encoded > > in, or should HTTP headers provide that information ? > > What is the most consistent wrt other resource types ? > > > The most consistent approach is to treat the headers > as definitive (if they are present), falling back > to the type information stated in the *link* (if > present). > > Of course, browsers are also "tolerant in what > they receive" and so they support things like > automatically inferring a mime type for a resource > based on the "file name [really, URL] extension" > or, yes, by sniffing contents (e.g., distinguishing > between "apparent HTML" and "apparent plain text"). > > So, browsers (user agents, processing agents, whatever > you want to call the category) sometimes sniff contents > and use other tricks but best practice is to use > the HTTP headers and the design of new standards should > assume best practices. > > -t > That is my understanding as well. So color me puzzled if different font formats and encodings should not be reflected in HTTP headers.
Received on Wednesday, 12 November 2008 01:11:52 UTC