Eli Zaretskii writes: >> From: Spencer Baugh >> Cc: 79201@debbugs.gnu.org, dmitry@gutov.dev, johnw@gnu.org, >> app-emacs-dev@janestreet.com >> Date: Mon, 11 Aug 2025 13:20:05 -0400 >> >> Yes, sorry, I missed that there are two separate mechanisms. I think >> set-process-thread should continue setting the thread member of the >> process object, but not the thread member of the fd_callback_info struct. > > Why not? set-process-thread can be called when the descriptors for > the process are already set up. That's unlike make-process, which > indeed doesn't set the .thread member of the descriptors. It's useless and causes this bug. >> Yes. Let's just make set-process-thread stop setting the thread member >> of the fd_callback_info struct. > > I don't agree it shouldn't see above. What problems do you see as the > result of that? It causes this bug. Here's an attached patch which makes this change. The thread member of fd_callback_info is then never set, so it can be entirely deleted. This doesn't need an announcement, because the behavior of "calling accept-process-output on a process locked to a different thread" still works the same way it always did. There's no (non-bug) user-visible change.