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