This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Enforce GMP/MPFR version requires and fix fortran/28276,27021
> If no one reviews the toplevel changes within the next few days, I
> will commit the changes under the implicitly approved rule.
The only time a configury patch may be considered "implicitly
approved" is if the patch is specific to something, and the maintainer
for that something has approved it. It still must be explicitly
approved, just not by the expected maintainer. We allow this because
usually, for example, target maintainers know more about configuring
their targets than the global build maintainers.
We're more strict with the toplevel files, because they're shared
across not just projects, but across repositories, and it's difficult
to tell what the other repository will need. Note that "happens to
only be used by <something>" is not the same as "specific to
<something>". GMP does not meet the specificity requirement, because
something in the other repository *could* use it. Disabling
libfortran for a specific target because it's unsupported, would.
Specifically, these are allowed:
* stuff inside a host- or target-specific clause of a case statement
may be approved by that host or target's maintainer(s).
* Building or not building a module (libfortran, libjava, etc) may be
approved by that module's maintainer(s). Likewise for listing build
dependencies for that module.
Historically, changing the build requirements wrt outside packages
(like texinfo or GMP) has always required discussion and explicit
approval.
Note also that all changes to toplevel files (and top's config/) must
be applied to both the gcc and the src (binutils/gdb/newlib/etc)
repository, such that they stay in sync. If you are not able to
commit to both repositories, you have to get a build maintainer's
approval anyway, because they'll be committing for you.