From: Martin B. <bl...@fu...> - 2006-03-30 06:07:28
|
On 3/29/06, Ben Finney <big...@be...> wrote: > "G. Milde" <g....@we...> writes: > > Something that can: > > - run as a pipeline (LaTeX on stdin, PDF on stdout) by default > > - allow any specified path for input file and/or output file > > - hide all the cruft that LaTeX processing creates, leaving no extra > files unless asked > > - properly handle Unicode, standard fonts, standard filenames and > locations, and other conventions that seem poorly supported by > LaTeX > > - ensure that the process ends with either an unambiguous error > state, or a complete PDF rendering of the input > > Currently pdflatex does *none* of these, and is thus a bad tool > (largely because it fails to sufficiently abstract the underlying bad > tools). I totally agree. I even started writing such a tool, I just want to be able to say rst2pdf.py <myfile.tex> and only see <myfile.pdf> as output. That's it, no crap, nothing else generated. I wrote the tool, copying my file to a temporary directory, and its dependent files. I ran into a bunch of issues with file inputs/includes from LaTeX, and then I got derailed onto some other open source stuff, and never finished it. When I finish it, I will simply delete the temporary files in the source location. Also, the tool must be able to plop some custom input into the generated LaTeX file. I always find myself editing it by hand when I want to make a "nice" document. Another avenue that is on my todo list is a nroff writer. That would do the job for a lot of simple documents, and produce a different-looking output from that of LaTeX. Another issue: I talked to Dave about it, multiple times, and he's quite hung-up on not including this tool in the standard docutils tools, for some reason (I can't remember which). Dave? |