> (b) you are not including the right sgml declaration for nsgmls to process
> the file as xml
Apparently, based on my success at the command-line level with Karl's
instructions, this is the problem.
> If you're using the tei-emacs installation as currently packaged and have
> installed it as we deliver it.
I must confess that I have not tried to install tei-emacs. I am a
Linux newbie, and much newer to Emacs. During my first six months of
trying to use Emacs as it came with Mandrake, I could not get either
SGML or Mule-UCS to function properly (I found out eventually that
there is a conflict between nsgmls and files that Mandrake uses to
process XML). I tried a number of times to reinstall both Emacs and
Xemacs from the tar.gz files, and could never get them working.
Then, to my great delight, when I recently switched to Red Hat 7.3,
the Emacs RPM that came with the installation set up both SGML mode
and Mule almost perfectly. So since then, I have been reluctant to
mess with any new installations of Emacs, unless it comes out as an
Let me just ask one basic question about tei-emacs for now: If I were
going to try to install it, would it be the case that I would
uninstall my present Emacs setup, download the tei-emacs.tar.gz, and
replace it with this package? If so, that's scary.
Thanks for your patience,
> then the only way I can see of getting the
> symptoms you describe is to give the system an SGML file to validate, or
> an XML file which invokes an SGML dtd (the latter is easily done by, for
> example, not specifying %TEI.XML as "INCLUDE"). Hence my original
> If however you're not using tei-emacs exactly as packaged, for example by
> supplying your own SGML declaration or not supplying one at all, then
> obviously the above hypothesis is entirely invalid, and I apologise for
> even formulating it!
> nsgmls comes with an SGML declaration that fools it into validating XML
> files. You will know if you are using this because itq generates a warning
> message to the effect that you supplied an SGML declaration. Why does it
> do this? Because XML files are not supposed to require an SGML
> declaration: they are not permitted to vary the reference concrete syntax.
> But nsgmls requires one because it is an SGML parser, not an XML parser.
> So if you supply one, it conscientiously warns you. This is one of the
> many reasons why some people on this planet seriously believe James Clark
> to be up there with Eric Clapton in the pantheon of contemporary
> On Tue, 20 Aug 2002, Charles Muller wrote:
> > Lou wrote:
> > lb> Tim is quite right. The XML entities are intended for use with XML
> > lb> files, not SGML. Hence their name.
> > I know that. I'm doing XML (using TEIXLITE.DTD), not SGML (please see my
> > original message). I thought that Tim understood that as well, and was
> > agreeing with me that Emacs/psgml is mistakenly validating XML files as if
> > they were SGML, at least concerning these external entity references.
> > lb> You say the files are OK with other parsers: which other parsers?
> > lb> and how OK?
> > I mean that the other parsers (other XML editors such as epcEdit, oXygen,
> > XMLSpy, Excelon) take the document in question to be valid, and do not
> > generate these entity reference errors. I was guessing that there might be a
> > setting in the psgml.elc file, but haven't been able to turn anything up as
> > yet. I do have a line in my .emacs file to tell psgml to read my files as
> > XML rather than SGML, but that does not seem to be working for this
> > particular instance.
Charles Muller <[log in to unmask]>
Faculty of Humanities, Toyo Gakuen University
Digital Dictionary of Buddhism and CJKV-English Dictionary
Mobile Phone: 090-9310-1787