On Tue, Feb 13, 2024 at 02:43:39PM -0500, Marek Polacek wrote: > On Tue, Feb 13, 2024 at 08:38:18PM +0100, Jakub Jelinek wrote: > > On Tue, Feb 13, 2024 at 02:24:11PM -0500, Marek Polacek wrote: > > > Sadly, I must admit this is looking like GCC 15 material. > > > > If deferred for GCC 15, can't we at least do some minimal > > change and just guard the member pedwarn with cxx_dialect < something? > > I could do something like that, but... > > > Given that -Wextra-semi isn't on by default nor included in > > -Wall -W, I think even accepting this for GCC 14 wouldn't be that > > risky. > > ...I also don't think it's that risky but technically, it's not > a regression I think. > > > Jason's decision. > > > > + /* If -Wextra-semi wasn't specified, warn only when -pedantic is in > > > > > > + effect in C++11 and below. DR 1693 added "empty-declaration" to > > the > > > > + syntax for "member-declaration". */ > > > > > > + else if (pedantic && cxx_dialect < cxx14) > > > > > > > > If it was a DR, did it apply just to C++14 or changed C++11 as well? > > It's got Status: C++14 so I thought that C++11/C++98 had not been > adjusted.
...but then so was DR 1601 which I implemented in https://gcc.gnu.org/git/?p=gcc.git&a=commit;h=e295e3d981355c61b72eca2ee58864958655cc31 and made it apply to C++11 as well. So I'm starting to lean towards adjusting this patch to make DR 1693 apply to C++11 as well, but not C++98. Probably not a very important distinction in practice though. Marek
