On Sun, Jun 07, 2026 at 09:14:02PM +0200, Patrice Dumas wrote: > Hello, > > I made two changes to the SWIG interface: > * it is now C only, no more Perl code call > * it is statically linked against the texinfo libraries > > It should now work on every platform. I changed the configure such that > if SWIG is detected, the interfaces are built and installed in the > default case. > > Does it look good?
That sounds alright, as long as there are proper checks for support and this doesn't break the build on previously supported platforms. I have tried building on my computer and it seems fine. The result in tta/config.log is that SWIG wasn't supported and so it is skipped. "make check" completed successfully. configure:49326: checking for swig configure:49365: result: no configure:49326: checking for swig3.0 configure:49365: result: no configure:49326: checking for swig2.0 configure:49365: result: no configure:49460: checking SWIG interfaces configure:49462: result: no > Note that since it is now C only, it means that it will depend more on > the underlying C library. It should not make much difference, except > for translations that are done right after parsing, they won't be done > if the libc does not handle LANGUAGE, as is the case for musl. When does this make a difference? As far as I understand, the SWIG interface is for user extension code, so nothing should break with texi2any itself, right? > There is now a failure in opencsw in the SWIG interface testsuite for > solaris 10, which happen to have swig installed, for one test. > Other than that, all the github CI pass, with swig installed when > possible. The Solaris 10 failure is a problem as it shows that the SWIG code may not be reliable if it is enabled. Will you be able to investigate the failure?
