On Tue, Jul 4, 2023 at 2:52 AM Tristan Partin <tristan(at)neon(dot)tech> wrote:
> The patch looks good to me as well. Happy to rebase my other patch on
> this one.
Thanks. Here is a slightly tidier version. It passes on CI[1]
including the optional extra MinGW64/Meson task, and the
MinGW64/autoconf configure+build that is in the SanityCheck task.
There are two questions I'm hoping to get feedback on: (1) I believe
that defining HAVE_MBSTOWCS_L etc in win32_port.h is the best idea
because that is also where we define mbstowcs_l() etc. Does that make
sense? (2) IIRC, ye olde Solution.pm system might break if I were to
completely remove HAVE_MBSTOWCS_L and HAVE_WCSTOMBS_L from Solution.pm
(there must be a check somewhere that compares it with pg_config.h.in
or something like that), but it would also break if I defined them as
1 there (macro redefinition). Will undef in Solution.pm be
acceptable (ie define nothing to avoid redefinition, but side-step the
sanity check)? It's a bit of a kludge, but IIRC we're dropping that
3rd build system in 17 so maybe that's OK? (Not tested as I don't
have Windows and CI doesn't test Solution.pm, so I'd be grateful if
someone who has Windows/Solution.pm setup could try this.)
[1] https://cirrus-ci.com/build/5298278007308288