tds 0.100

Paul Vojta TWG-TDS@SHSU.edu
Tue, 17 Oct 1995 12:14:28 -0700 (PDT)


On Tue, 17 Oct 95 09:49:18 BST, David Carlisle <carlisle@cs.man.ac.uk> wrote:

> >>>>> Paul Vojta writes:
> 
> > May I suggest the following diff:
> 
> [omitted]
> 
> Sorry for coming to this a bit late, but I still do not really
> understand this proposal, with either wording.
> 
> Is the intention that TeX supports \input latex/file.tex to mean to
> find file.tex in some sub-directory recursively below latex for
> instance  \input latex/graphics.sty finds latex/graphics/graphics.sty

Yes.  Specifically, if you say \input latex/file.tex, and the search path
is .:/usr/local/texmf/tex//, then:

   (1)	first it checks for ./latex/file.tex
   (2)	then it looks recursively within /usr/local/texmf/tex/latex for a file
	named file.tex.

> If not, and  \input latex/file.tex  is supposed to mean that file.tex
> is in a directory called latex, then the TDS would need completely
> restructuring, not just relaxing a uniqueness restriction.

Not applicable.

> Furthermore, the particular example |\input latex2e/file.tex| is a bad
> one as the TDS directory name for LaTeX is latex not latex2e,

Correct; mea culpa.

> and the
> primitive TeX \input syntax (without {}) is strongly discouraged in
> LaTeX.

Of course in LaTeX you use \documentclass or \usepackage, but with the
LaTeX macro package that eventually reduces to \input.  My intent is that
the macro package would insert "latex/" or whatever; hence the wording
``(with the cooperation of the macro packages).''  See my message of 24 June.

Also, on Tue Oct 17 08:04:46 1995, "K. Berry" <kb@cs.umb.edu> wrote:

>     > ! However, this is not currently practical, since no current {\TeX}
>     > ! implementation supports this.
> 
> Ohh, you're speculating on a new syntax for filename searches within TeX ...?
> I didn't understand that before.

More precisely, an extension of the current syntax; see above.

>     > In the current wording, it's not at all clear to me how the TDS tree
>     > might look different.
> 
> Probably because it wasn't clear to me, either. What is your conception
> of how the tree would change?

My conception is that it wouldn't change.

>     Is the intention that TeX supports \input latex/file.tex to mean to
>     find file.tex in some sub-directory recursively below latex for
> 
> I don't know. Paul, is that what you had in mind?
> If so, I don't think this is the right syntax to contemplate;
> latex/file.tex should mean look for a file file.tex in a directory latex,
> because that's a regular filename path.
> Perhaps //?

Hm.  I hadn't considered that.  I figured the // would stick around from the
TEXFONTS variable.  Actually, I still prefer only /.  To use LaTeX as an
example, with //, latex would translate \usepackage{foo} into
\input latex//foo.sty, but then \usepackage{ams/foo} would translate into
\input latex//ams/foo.sty.  It would work, but it wouldn't be pretty.
Also, if '.' is at the beginning of the search path, and if a directory
./latex exists, then it would do an unintended recursive search within that
directory.

>     Furthermore, the particular example |\input latex2e/file.tex| is a bad
>     one as the TDS directory name for LaTeX is latex not latex2e, and the
>     primitive TeX \input syntax (without {}) is strongly discouraged in
>     LaTeX.
> 
> Thanks for catching that. I guess the example will be
> plain//testfont.tex or something.

plain/testfont.tex is fine with me.

--Paul Vojta, vojta@math.berkeley.edu