On Mon, 2023-01-02 at 10:47 -0500, Dmitry Goncharov wrote: > On Mon, Jan 2, 2023 at 9:50 AM Paul D. Smith > <invalid.nore...@gnu.org> wrote: > > I decided to own this glob implementation > > Paul, do you mean that you decided to maintain glob impl in make? > If yes, do you still want to link against glibc impl, when available?
Yes GNU Make will still use glibc version of glob when it's available. All I'm doing is cleaning up some of the really old / unnecessary stuff in the local glob/fnmatch implementations that ship with GNU make (removing support for pre-ANSI C function declarations, removing deprecated "register" keyword, assuming C headers like stdint.h and stdlib.h and that they correctly declare C functions like malloc, realloc, etc.) I don't plan to change (for example) the configure.ac code that decides when to use the local vs. system implementation. Previously I had tabled bug reports for those files since they weren't "owned" by GNU make, but that doesn't seem tenable anymore since I can't really envision being able to adopt the gnulib implementation of glob/fnmatch.
