Enrico Forestieri wrote:
On Mon, Apr 21, 2008 at 09:44:35PM -0400, rgheck wrote:
Enrico Forestieri wrote:
[...]
What about child documents in the approach you are pursuing?
This is the hard case. The thought I had was simply that child documents
would appear as elements of the bundle, treated no differently from
anything else. Of course, they would themselves be bundles. It's here
that you can really see the different feel of this approach, perhaps.
Note that the script, other than addressing child documents, also catches
files inputed through ERT and also examines them for further inclusions.
If you input foo.pstex_t, the script also catches foo.pstex. If you use
\usepackage{foo}, the script examines foo.sty (but only if it is in the
document directory) to see whether there are other files that should be
included. For this reason I said that maybe it is more complete.
Right, that may be so. I'm certainly less ambitious than you are here.
It's a different goal, in some ways.
Are we including foo.sty in the bundle only if it's in the document
directory? I wouldn't think we would want it if it's in the main TeX
tree. But what if it's in my local tree?
I've got some other questions about the script and how it works, but we
can discuss those later. Here are a couple. First: It looks as if in
many cases, since bst files and the like are being bundled, that in fact
the lowest common directory will be the root directory. In that case,
the archive is encoding absolute filenames rather than relative ones,
and Jose, at least, has expressed concern about the security
implications of that. This can probably be addressed somehow, perhaps by
not bundling anything the user doesn't own? (Well, maybe that doesn't
work so well on Windows.) Second: I'm not sure I understand the
reversibility aspect of this. Is the idea that the archive can simply be
unpacked via tar -zxvf? If so, then maybe that's OK. Only "experienced
users" would want to do that. But that does make reversibility hard, and
I'm still worried about any LyX-like file that could in principle write
to locations where LyX might seek to find scripts. This seems like
potential trouble.
I hope our conversation about this can be less difficult than the last
one. I'm not trying to be difficult, though maybe I'm nonetheless
succeeding. ;-)
rh
