Re: [PATCH 0/5] libc++ update to 01f3a59fb3e2542fce74c768718f594d0debd0da

[Joerg Sonnenberger <joerg%bec.de@localhost>]

On Tue, May 12, 2020 at 02:32:11PM +0200, Kamil Rytarowski wrote:
> On 10.03.2020 13:58, Kamil Rytarowski wrote:
> > On 09.03.2020 18:03, Joerg Sonnenberger wrote:
> >> On Mon, Mar 09, 2020 at 01:08:39PM +0100, Michał Górny wrote:
> >>> Here's a series of patches to update our libc++ to the version
> >>> corresponding to our current LLVM/Clang.  I've tried to grasp the idea
> >>> behind our import, and follow suit, avoiding adding unnecessary files
> >>> and installing unnecessary system header overrides.
> >>
> >> Don't waste your time on it. I've been dealing with it for a while and
> >> it is not enough by itself, even ignoring the max_align_t case.
> >>
> >> Joerg
> >>
> > 
> > Please list what is missing.
> > 
> > If that is math.h compatibility, we use math.h from libc++ wrapping our
> > native math.h.
> > 
> 
> Ping? Old libc++ affects our two GSoC tasks (*). Can we get it merged soon?
> 
> (8) lib/54955: basesystem's libc++ is too old and contains fixed
> upstream (crashy) Undefined Behavior

The in-progress changes can be found together with the pending LLVM
update on anonhg. I'm still going over the list of regressions in pkgsrc
to see what is bad software and what are issues in libc++, which has
been slower than necessary due to all the other unrelated breakage, but
the list has become moderately short now.

Joerg