Wed, 11 Jun 1997 08:23:58 -0400
Sebastian Rahtz's message of Wed, 11 Jun 1997 10:29:02 +0100
> what about those of us who convert the LaTeX to SGML, preserving the
> concept of automatic numbering? what you suggest would be seriously
> retrograde, IMHO
Only if you believe that element numbers are not data. It is not SGML
that dictates the numbers cannot be data, it is your DTD. There seems to
be a prevalent misconception in the SGML community that the numbers are
an inessential part of the document, and that it makes no difference if
numbering style is changed from 1 to i to a. This may be practically
true in many cases, but the kind of material in which it is true least
often is mathematical material, which happens to be the stuff that my
employer specializes in---hence my colored view of the situation.
In any case if explicit numbers were added to LaTeX 3 documents I don't
see that it would impair your current approach any. Just have your
LaTeX2SGML converter discard the numbers.
> > I am inclined to think, therefore, that the ideal submission process
> > should involve a step where all the automatic numbers are replaced by
> > their explicit values from the .aux file. (In particular, what the
> also, the number still needs to be abstract. i don't want an explicit
> `1.1' when the style is `I a', for instance!
Actually a full number spec should include both value and formatting,
separated, for maximum flexibility. I have some prototypes lying around
That the numbers can be generated automatically while an author is
working on a document does not imply that they can be omitted or changed
in form in the document that readers ultimately see. For example:
---A mathematician who chooses to number list items with Greek
letters will find a change to standard numbering by the publisher
unpalatable; it destroys the nuances of the original.
---In a mathematical document where \bullet or \square is used as a math
symbol the author will naturally tend to shun the use of the same symbol
for itemized lists. Putting the document through a publishing process
that ignores this constraint might be a significant disservice to the