Ross Boylan <[EMAIL PROTECTED]> wrote: > First, on RTFM: why would a regular user even think to look at > geometry?
Because he's learned that additional packages are the only way to get properly sized PS and PDF files, and geometry is one of them? > Second, I agree that specifying a particular driver is ugly; it's > just that at the moment that seems to work. Yes, and I hope that we'll be able to change this so that it isn't needed to specify a driver. > Third, we've mostly been talking about geometry, a LaTeX package. > What is a regular TeX user to do? If he cares about document-specific papersizes, he can use the \special command just as the latex packages do. Or he can resort to using texconfig-sys or specifying the papersize in the dvips commandline. >> But I don't see how this invalidates the argument that a system-wide >> default paper makes no sense. In most cases, a per-document paper >> setting does make sense, and the special cases are, well, special cases >> which don't speak for a system-wide default. > > My point was two-fold. First, that a real solution to this problem > will need to include a capability that is currently lacking, namely > getting the page size from the tex source to the post-processing tools > on a per page basis. Second, if the argument that "usually all pages > in a document have the same page size" is good enough to argue for a > document-wide default page size, why isn't the argument that "usually > different documents have the same page size" good enough to argue for > a system-wide default page size? I agree with your first argument. As for the second: The statement "The proper solution is to specify paper size(s) in the document" is true in particular because of portability issues: I believe it should be possible to transfer a (La/Con)TeX document to an other machine with an other TeX system and still get the same typeset output. On the other hand, it's true that I use the same paper size most of the time, and therefore it also makes some sense to have a system-wide default: See the latest progress in #402994. But when one makes use of this feature, they should be aware of the limitations. > More basically, the point that documents may have different page sizes > does not imply that a system-wide default makes no sense. From my > perspective, it makes perfect sense. It is good to be able to specify > a page size; it is also good to get a sensible page size when you > don't specify one. (And, to return to the origins of this bug, it's > even better on Debian if that default comes from /etc/papersize). Yep. Please note that we've not tagged #402994 "wontfix" as all the earlier libpaper bugs... >> As I said, I hope future LaTeX versions will do that (maybe >> ConTeXt does it already?). >> > I take it that modifying LaTeX to write out the specials (as is > currently done by geometry) is out of the question? It would be nice > if \documentclass[papersize]{xxxx} were enough to get the necessary > info to the post-processing tools. Yes - I hope this is going to happen with LaTeX 3.0. But not earlier, since it's a change that somewhat breaks compatibility and will not happen in LaTeX 2e. And the release of LaTeX 3.0 is, well, *very* undefined. Something like "not this decade, maybe next". Regards, Frank -- Dr. Frank Küster Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich Debian Developer (teTeX/TeXLive)