David Carlisle <[log in to unmask]> writes:
: I think the problem is is that SGML has been around since whenever it
: was, about the same vintage as TeX (older if you count GML), and with
: that power it had essentially zero impact on mainstream applications.
: (Although of course it was used in house in many larger companies.)
One can speculate about why SGML has been such a well kept secret. It
takes some work to appreciate it. That does not means that it does
not work nor that there are not reliable free tools nor that it would
not be fairly easy once we come to understand what it can do for us to
generate tools that are optimized for our purposes.
: So it seems to be the case that the `SGML experience' says that as soon
: as you want to move the document anywhere then it should be in a
: standard concrete syntax, with all minimisation normalised. ie be XML.
Probably. This is "public exchange".
: Of course how one personally authors the document is up to the author,
: it might be gellmu, or you might use latex and then tex4ht or omega-mml
: or frame+SGML or MSOffice2000 or whatever you want.
: If however you expect to allow people to edit the transmitted document
: and send you back the changed version (eg a co author, or a journal
: editor) it makes sense to use a system that really uses the transmitted
: XML form of the document, rather than a gellmu or latex->*ml style
: translation where it is not necessarily easy to go back.
: Within the tex world it makes sense to pass tex documents,
(Two LaTeX co-authors should exchange drafts in plain TeX?)
Anyway, this is "shared authoring".
(contin. from above)
: but I do not
: think it makes sense now to be passing SGML documents that make
: essential use of a non standard SGML declaration, and SGML minimisation
: features. The number of XML applications appearing increases at a
: ridiculous rate, the number of applications that can actually handle
: minimised SGML documents is small and static, I fear.
Note: HTML 2.0, HTML 3.2, and HTML 4 all use different SGML declarations,
none the default.
For this purpose one should perhaps view an SGML document type as a
decl/dtd pair. Of course, there is no decl for an XML.
There are many more web readers than math authors. Math authors have
done very well for many years with a very small number of tools that
are not used by the masses.
(1) Authoring, (2) shared authoring, (3) public exchange, (4) private
archiving, and (5) public archiving are different things with different
Obviously, to the extent that it is sensible to adopt different formats
for different purposes, it is desirable to have automatic processing to
faciliate conversions. Many such conversions should be fairly easy.