Re: LINK TYPE=override/type

Liam Quinn (liam@htmlhelp.com)
Fri, 23 Jan 1998 08:36:25 -0500


Message-Id: <3.0.5.32.19980123083625.0096f700@undergrad.math.uwaterloo.ca>
Date: Fri, 23 Jan 1998 08:36:25 -0500
To: www-html@w3.org
From: Liam Quinn <liam@htmlhelp.com>
In-Reply-To: <Pine.SUN.3.96.980123010746.5904A-100000@galileo.cris.com>
Subject: Re: LINK TYPE=override/type

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

At 01:16 AM 23/01/98 -0500, Bill Bereza wrote:
>On Thu, 22 Jan 1998, Ian Graham wrote:
>
>> This assumes that you can self-consistently manage all linked 
documents,
>> which is not the case.  For example, someone linking to your document 
might 
>> specify the type of your document using TYPE. But, if you subseqently 
>> change the type, their documents will break.   
>> 
>I think that if you're going to be changing the MIME type of documents
>you're bound to have problems whether or not someone was using TYPE.

Why?

>> At the same time, I believe TYPE should play a role in error handling 
>> -- for example, if the HTTP-specified type seems to be in error, then 
>> the browser could fall back to the TYPE-specified value, and see if 
that
>> works. And if that doesn't work, the browser could try heuristics on 
the
>> file, to try and determine the correct type (i.e. ... guess!)
>> 
>
>No, a browser can not ever do this, because there is no way for a
>browser to know or guess if a server is mis-configured. Any browser that
>tries something like that is going to mess things up very much.

In general, yes, but there are cases in which guessing can be more 
appropriate than simply ignoring the badly served file.  For example, if 
foo.css is linked with <LINK REL=StyleSheet HREF="foo.css" 
TYPE="text/css"> and the server returns a Content-Type that is definitely 
not a style sheet (e.g., text/plain or application/octet-stream), then I 
think that using the TYPE attribute would be more appropriate than 
ignoring the file.  However, if <A HREF="foo.css" TYPE="text/css"> is used 
as a link, then I think browsers should trust the served Content-Type, 
whatever it may be.

>Either the browser should always use the HTTP MIME type or it should
>always use some author-specified TYPE value. A browser should not be
>using the server's type value in certain cases while ignoring it if it
>"thinks" the server is in error. According to HTTP we must assume that
>the server is never in error. 

Right, and in general I don't think TYPE should be used as an override.  
The LINK example I describe above is really a special case of error 
recovery on the part of the browser, and could proceed in much the same 
way without the TYPE attribute.

TYPE is most useful, IMO, for non-HTTP links (particularly FTP), and for 
linking style sheets and scripts to allow non-supporting user agents to 
avoid fetching the resource.  Creative user agents (or authors through 
style sheets) could also use TYPE to distinguish different kinds of links, 
perhaps with an icon for different media types.

-----BEGIN PGP SIGNATURE-----
Version: PGP for Personal Privacy 5.0
Charset: noconv

iQA/AwUBNMic2PP8EtNrypTwEQKvlwCgsDKCejruNb63Pd6v4TCfcVDLQiYAniGs
eBP7gRiFrlhZqe4yX0Ajk342
=q/MV
-----END PGP SIGNATURE-----

--
Liam Quinn
Web Design Group            Enhanced Designs, Web Site Development
http://www.htmlhelp.com/    http://enhanced-designs.com/