On Mon, 27 Jun 2022 at 21:40, Yann Ylavic <ylavic....@gmail.com> wrote:
> On Mon, Jun 27, 2022 at 8:35 PM Ivan Zhakov <i...@visualsvn.com> wrote: > > > > On Mon, 27 Jun 2022 at 21:14, Yann Ylavic <ylavic....@gmail.com> wrote: > >> > >> On Mon, Jun 27, 2022 at 8:08 PM Ivan Zhakov <i...@visualsvn.com> wrote: > >> > > >> > On Mon, 27 Jun 2022 at 21:01, <yla...@apache.org> wrote: > >> >> > >> >> +#ifdef WIN32 > >> >> +typedef apr_status_t (__stdcall *encdec_fn)(char*, const char*, > apr_ssize_t, int, apr_size_t*); > >> >> +#else > >> >> +typedef apr_status_t (*encdec_fn)(char*, const char*, apr_ssize_t, > int, apr_size_t*); > >> >> +#endif > >> >> > >> > This uses assumptions about calling convention used on the Win32 > platform. I don't think it's a good thing. Do we really need an array of > callbacks? May just inline all these calls? > >> > >> APR_DECLARE() seems to always __stdcall, as opposed to > >> APR_DECLARE_NONSTD()'s __cdecl. > >> But I barely understand these Windows subtleties, so yeah I can inline.. > >> > > > > Yes, *currently* it is always a stdcall, but it could change so it's > better do not have such assumptions. > > r1902286. > > Looks good! Thanks! -- Ivan Zhakov
