Hi Bruno, Bruno Haible <[email protected]> writes:
> Hi Collin, > >> Using the reproduction steps Tim shared [1], this patch fixes the >> redeclaration on MinGW: >> ... >> >> I'm not sure if that is correct and/or if it should be documented. So I >> will hold off pushing for review. > > Thanks for the effort. But no, this patch does not go into the direction > that the stakeholders (GNU emacs [1], GNU gettext, GNU libunistring) want. > These packages want to use mingw _without_ libwinpthreads, because the > threading primitives in this library are buggy [2][3]. Currently 'nanosleep' > is the only import of these packages from libwinpthreads. And while > this 'nanosleep' implementation is not buggy, we are trying to get rid > of the dependency to libwinpthreads altogether; it would simplify > the packaging of native Windows binaries (as done by Corwin for Emacs > or by Michele Locati for gettext). > > [1] https://lists.gnu.org/archive/html/emacs-devel/2023-08/msg00635.html > [2] https://lists.gnu.org/archive/html/bug-gnulib/2023-04/msg00176.html > [3] https://lists.gnu.org/archive/html/bug-gnulib/2023-08/msg00092.html Thanks for the review. So would the correct fix to be to change the cross guess for "mingw*" to something other than "guessing no"? That would avoid linking to winpthreads and define nanosleep to rpl_nanosleep, I think. Collin
