On 15 January 2018 at 17:43, Mathieu Lirzin <m...@gnu.org> wrote: > Sorry if I am opening a can of worm, but what about moving the main > development repository from Subversion to Git? :-) > > We could ask Eric Raymond to help with the migration if he has some time > for it. He has already helped other GNU projects such as Emacs and > Groff to migrate to Git with a specialized tool ‘reposurgeon’ which he > maintains. > > WDYT?
>From my point of view this doesn't add much as I am comfortable using SVN. I have not been doing much work on Texinfo in the last few months (partly due to lack of time, partly due to lack of motivation). If someone else was doing more work I would be happy to follow their choice if they wanted to transfer to using Git instead. Here's what I have in mind for future work on Texinfo (all off the top of my head): * Make texi2any faster? I did a quite some work rewriting part of it in C but I did not finish it. (Since then, I made a few changes to the Perl code making the work in C incompatible. The C code could be updated but this would take some work.) I expect there would be lots of subtle incompatibilities between the C and Perl parser code. However, the main issue is that the program would still not be as fast as the original makeinfo (written all in C). The part of the code that recurses over the parse tree and outputs the output file is a large part of the execution time and I couldn't see any way to make it faster. Maybe a challenge for a Perl expert? * Other internal improvements to texi2any, e.g. for better error detection or to make writing a new back end easier. * Automatic @node for section commands * LaTeX output for better output customisation. Better output customization in general. * Test suite for texi2dvi (nice to have but not vital) * Integrate HTML/JavaScript web user interface code into texi2any and the Texinfo distribution.
