https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120353
--- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> --- (In reply to qinzhao from comment #6) > (In reply to Jakub Jelinek from comment #5) > > Why has this been backported to 15 branch? > > Generally accepts-invalid bugs aren't backported to release branches, or > > similar changes which cause rejection of more code than was accepted before > > on the branch, or more warnings than before on the branch. > > It is ok to e.g. error on something we'd ICEd on before. > > The reason for this is so that people need to do (if possible) porting of > > their codes to a newer gcc version just once a year for major releases, so > > that they don't need to repeat it every week or month again. > > So, you mean, for a patch that might make the compiler issue more warnings > should not be backported to a release branch? There can always be exceptions if there is a strong reason, but yes, only new major releases should be generally more strict in what they accept. Furthermore, the normal release branch rules are that only regression bugfixes and documentation fixes are backported. Again, exceptions are possible, but shouldn't be done frequently. This is neither a regression (at least is not marked as such) and even a regression fix would be strongly preferred not to introduce extra diagnostics which wasn't there in 15.1. > since such new warnings in a > previously released compiler might force the users to upgrade to newer gcc > version sooner than they want. > > should I revert the patch from GCC15? I'd say so, unless other RMs disagree.
