> Well, to be honest, it now looks to me like a design mistake. The
> idea was to provide a simple mechanism to allow users to add new
> entity and notation declarations to TEI Lite. Only after we had
> published the DTD did someone point out that it was an exceptionally
> poor idea to use a public identifier to refer to a file which was, by
> nature, intended to vary from site to site.
I cannot accept Michael's apology because I don't think the use of a
public identifier in this case was necessarily a mistake. I like the
idea of having a standardized place to store user modifications,
essentially a guideline for saying that if people are going to modify
the TEI Lite DTD, they may want to do it in a standardized way.
While some modifications may be site-specific (e.g., display entity
sets), others may be project oriented: users may modify the TEI Lite DTD
not to address peculiarities of their site, but to address the
peculiarities of their encoding project (e.g., new elements and
attributes). These extensions might reasonably need to be part of the
document as it is exchanged.
> And since, in practice, very few users ever seem to want to extend TEI
> Lite in the way allowed by this mechanism, it seems to be much more
> trouble than it's worth.
I think it's worth leaving in as long as the default for most users
(doing nothing) doesn't result in parser error messages.
> So in the next version of TEI Lite, the usrmods file should go away.
I'd propose keeping it, but including in the distribution the dummy file
that Michael has configured on his own machine. Most users won't know
the difference, but those who want to customize the TEI Lite DTD will be
able to do so by editing that particular file.
> Yours in abject penitence,
Off your knees and sin no more ...
Professor David J. Birnbaum email: [log in to unmask]
Department of Slavic Languages url: http://clover.slavic.pitt.edu/~djb/
1417 Cathedral of Learning voice: 1-412-624-5712
University of Pittsburgh fax: 1-412-624-9714
Pittsburgh, PA 15260 USA