Message-Id: <9211231316.AA15699@pixel.convex.com> To: Dave_Raggett <dsr@hplb.hpl.hp.com> Cc: timbl@nxoc01.cern.ch, www-talk@nxoc01.cern.ch Subject: Re: Supporting the Book metaphor In-Reply-To: Your message of "Mon, 23 Nov 92 11:23:46 GMT." <9211231123.AA12237@manuel.hpl.hp.com> Date: Mon, 23 Nov 92 07:16:20 CST From: Dan Connolly <connolly@pixel.convex.com> >Pictures >======== Answer: MIME. I coded a prototype replacement for HTParseFormat that groks several mime formats, including text/plain, text/x-html, message/rfc-822, message/external-body, and multipart/mixed. We just need to add plug-and-play modules for other MIME types like image/gif, image/x-tiff, etc. More on that after I get some sleep and write it up... >The Book Metaphor >================= > Answer: The DocBook DTD from O'Reilly and HaL. It's an SGML DTD, but it's not as simple as HTML. We need to get serious about supporting SGML. I've spent the whole last week with my nose in the SGML standard. I think I've got a handle on it now. I've written some code to do the low-level reading of SGML. It's reentrant, doesn't use malloc(), and completely conforms to the SGML standard (except for marked sections and wierd newline conventions). I even added a couple work-arounds for the problems with existing HTML data. And -- I integrated it into the MidasWWW browser. Works great. I hope to sync up with the author soon. I'm looking at integrating this SGML_read library into the linemode browser, but the linemode code isn't reentrant, and that makes life hard. What other HTML parsing engines do we have out there? Tim: you said something about the NeXT editor being hard to fix. Why is that? It looks like there are three lines in HTParse.h that write the HREF=foo attributes. I'll send you patches if you like (though I can't test them -- no NeXT!) Dan