https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121486
Jeffrey A. Law <law at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |WAITING
Last reconfirmed| |2025-11-11
Ever confirmed|0 |1
--- Comment #2 from Jeffrey A. Law <law at gcc dot gnu.org> ---
So we discussed this during the patchwork call today.
In general the expectation is that version numbers for the ISA and extensions
are not supposed to be meaningful going forward from RVI. If there is a
behavior change in a specification, then that's supposed to be a new extension
rather than a new version of an existing specification.
The notable exception here are the some changes in the F/D extensions that
happened a few years back, but the belief is no cores were ever produced with
the older version of the F/D specification. The other exception is the 0.7 V
extension which is now known as the thead-vector extension.
With that background the consensus was that we should accept version numbers,
but issue a soft diagnostic that they are ignored. This would be true for both
the --with-arch and -march strings (and I'll be marking the -march variant of
this bz as a duplicate momentarily).
--
So the outstanding question here is what precisely are you trying to accomplish
using the version numbers in your --with-arch/-march strings? While we think
the overall direction is sound, there may be something you're trying to
accomplish that we're papering over.
