On Mon, 17 Mar 2025 at 10:24, Giuseppe D'Angelo <giuseppe.dang...@kdab.com> wrote: > > Hello, > > On 17/03/2025 10:03, Tomasz Kaminski wrote: > > It seems that the patch is causing *first to be moved twice into the > > same position. > > Good catch. I think I need to move it *back*, at least that's what > __uninitialized_construct_buf_dispatch seems to do.
I see that the version on forgejo moves it back with: *__first = std::move(__buf); OK for trunk with that change. > > Given that we have constexpr allocations now, do we really need to avoid > > allocating temporary buffers, and using very slow implementation? > > That's an option but I wanted to go there for GCC 16 (unless you > greenlight me that this is doable right now). I think waiting for GCC 16 makes sense. We should also replace the non-reserved name in _Temporary_buffer that I've just reported as PR 119496. > I've written some more ruminations on this issue here: > > https://forge.sourceware.org/gcc/gcc-TEST/pulls/44#issuecomment-792 > > > > PS: I've noticed that I have already accidentally bumped the FTM for > constexpr stable sort; that's a bug as these two last algorithms were > not inculded. I can decrement it for the time being, then rebump it again? With the inplace_merge and stable_partition patches approved, can you bump it again now? That would presumably resolve PR 119488.
