On Wed, Jun 12, 2024 at 05:59:14PM -0400, Peter Geoghegan wrote: > There is also one oddball case, not quite in either category. This > involves zic.c's declaration of > link(), when it should actually just be using the #include <unistd.h> > declaration. That's another weird upstream code thing -- this isn't > exactly fully under our control. I've avoided doing anything about > that, but perhaps I should have proposed a patch for that, too (it's > fairly similar to pg_bsd_indent). What do you think of that idea?
That one seems to be synchronized somewhat regularly, and I haven't been the one doing the synchronizing, so we might want to be a little more cautious there. But I do see a couple of commits that have touched it (e.g., 235c0f6, c4f8e89, 0245f8d). At a glance, it looks like the link() stuff might be intended for Windows. I see we have our own version in win32link.c, so your idea to remove it in favor of the unistd.h declaration seems like it ought to work. -- nathan
