On Fri, Jun 06, 2025 at 02:34:09PM +0200, FX Coudert wrote: > > In commit 5e918a4db9e4a5bdbeafec6881fa8b22a55d3789, regenerated files > were not included in the commit as they should have been. Therefore, a > whitespace fix was not propagated. Sync generated files now, as obtained > from a run with --enable-maintainer-mode. > > Tested on x86_64-linux, OK to push?
Yes. Thanks for the cleanup. > > Also, I’d like to follow-up on the discussion in the PR > (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116400). There is agreement > that the current situation is not stable, regardless of the current bug > (which means that generated files are written to the build tree, and not the > source tree). Basically, the --enable-maintainer-mode option is highjacked by > libgfortran to regenerate these source files, but it is not its purpose: it > should deal with autoconf/automake build files only. > > What other GCC libraries does instead is: they have a script that triggers > regeneration of source files, and that script is run manually by maintainers > when needed. I have proposed a simple script to do just that, and I intend to > improve it (allow customisation of which m4 program to call, for instance). > But before I spend more time there, I wanted to raise this issue to the list: > is there general agreement to proceed in this direction? > Yes, I think this is what we should do. When I use --enable-maintainer-mode on FreeBSD it touches a number files outside of libgfortran. As mentioned in the audit trail of PR116400, I tested your script and indeed needed to modify it to use gm4 (GNU m4) instead of m4 (FreeBSD m4). I haven't had time to integrate the small script posted by Jonathan to find a usable m4. It looks like he lifted that out of configure. Thanks for working on a fix. -- Steve
