- From: Keith M. Corbett <kmc@harlequin.com>
- Date: Wed, 17 Jul 1996 17:39:15 -0400
- To: "Lee Daniel Crocker" <lcrocker@calweb.com>, Pam Kagan <pam@safari.net>, www-style@w3.org
At 03:14 PM 7/17/96 -0400, Pam Kagan wrote: >I was under the impression that CSS tags would be invisible >to non-compliant browsers. Style markup can only be invisible if it is a) used in external style sheets, or b) used in the HEAD of the document, or c) buried in comments. Unfortunately there are problems with all three approaches: a) external style sheets are harder to use than embedded markup, and they don't work very well in MSIE beta 1, so HTML authors don't have much experience with this feature; b) Netscape displays text in the HEAD of documents (this is a bug); and c) putting style markup in comments causes problems for existing tools that throw away comments. A better design for STYLE would be trivial if browsers conformed to the SGML and HTML standards but... At 12:31 PM 7/17/96 -0700, Lee Daniel Crocker wrote: >The designer of this page should be take out and shot. As should the designers of the HTML parsers in all the popular browsers. This is my pet peeve against Web browsers: if they can ignore element tags they don't understand, such as <SCREEBLE ON MY SCREEN> why must they display the same tag with all the text if it starts with a question mark? <?SCREEBLE ON MY SCREEN> In SGML terms the tag beginning with ?SCREEBLE is a "processing instruction", or PI. PIs are intended for special or "out of band" (non SGML) markup, such as formatting instructions that can be interpreted by some but not all of the systems that can process the data. Too bad the browsers don't ignore PIs they don't understand. If they parsed SGML correctly, authors would be able to use CSS without hiding it in comments. There are no HTML tags in popular use that begin with a question mark, and handling PIs according to the SGML spec would be trivial, or would have been back before the Web took off. (Too late now I suppose.) CSS advocates have recommended using SGML comments as a workaround, but this causes the CSS style information to drop out of (most? all?) conforming SGML systems: <STYLE><!-- font-family: bizarro --></STYLE> I think authors would accept PIs readily. For example, this causes CSS to appear as text in pre-CSS browsers, which is clearly bad style: <STYLE>font-family: bizarro</STYLE> But the following is very clearly something different from text, namely a formatting instruction: <?STYLE CSS font-family: bizarro> /kmc
Received on Wednesday, 17 July 1996 17:42:22 UTC