On 13/07/2022 20:16, Jeff King wrote: > On Wed, Jul 13, 2022 at 03:42:38PM +0100, Adam Dinwoodie wrote: > > Hmm. So I had assumed that the problem was unlink()ing the socket path > while the client was still trying to read(). If that's the case, then I > _think_ the minimal reproduction below should also trigger the problem. > That might give you something more useful to show to Cygwin folks. Hmm, good find - I looked at this commit while searching the gitweb pages the other night, but didn't think it was relevant! :( Oh, well ... I don't have too much time tonight, but I gave your program a quick try: $ gcc -o test-unix-sock test-unix-sock.c $ ./test-unix-sock $ ls -l my* ls: cannot access 'my*': No such file or directory $ ./test-unix-sock $ echo $? 0 $ HTH. ATB, Ramsay Jones
