On 29 August 2015 at 12:40, Vincent Belaïche <[email protected]> wrote: > Some proposal below... > > Le 22/08/2015 00:26, Karl Berry a écrit : >> At this point, I guess I have to conclude it is better to use -recorder >> if it is available. >> >> k > > > Hello, > > Attached is a patch that solves the problem for me. It is a bit long, > because I did it as backward compatible as possible. > > By setting environnement variable TEXI2DVI_USE_RECORDER you can suppress > the additional prior testing and make this novel texi2dvi almost as fast as it > used to be. > > VBR, > Vincent Belaïche > > >
Here are notes I made while looking at this: I see no need for a command-line flag. "pushd" and "popd" are bash extensions, use cd instead. Eliminate duplication between the two generated_files_get functions. Remove unnecessary semi-colons. Can someone tell me why there is all this stuff in texi2dvi for processing LaTeX files? Possible problem: a *.fls file meets the *.??s glob pattern for sorted index files. Probably not a problem. Two-space indents. Space after \openout in regexp. Check for single quote after filename. Remove $ anchor from end of regexp. Format for openout lines must differ between web2c and MikTeX. Change yesmaybe to nomaybe as per suggestion. Remove env var from help message - there should be no need for the user to think about this. Move setting of catcode_special variable, to just before use of make_tex_cmd. But there's a -~ option, so check it hasn't been set. Renaming of functions and comments - matter of taste. We'll also need a ChangeLog entry. Here's the current state. It would benefit from more testing and review. I'l have another look at it in the next couple of days.
texi2dvi
Description: Binary data
texi2dvi.diff2
Description: Binary data
