On 23/07/2025 14:36, Eli Zaretskii wrote: >> Date: Wed, 23 Jul 2025 04:39:33 +0300 >> Cc: sbaugh@janestreet.com, 76969@debbugs.gnu.org >> From: Dmitry Gutov >> >> On 22/07/2025 15:54, Eli Zaretskii wrote: >>>> So if we make it an optional argument of 'make-thread', and the default >>>> is for thread's buffer to be killable, that might call for a name change >>>> (flipping the meaning). >>>> >>>> Should it be >>>> >>>> (make-thread FUNCTION &optional NAME BUFFER-PROTECTED) >>>> >>>> ? >>> I'd prefer to call that argument BUFFER-DISPOSITION. >> >> I had to google it just now. A term from POSIX signals? > > Or from email messages. In the context of emails the meaning seems different (more like how the content is displayed). Although it just occurred to me that the word is derived from the root "dispose". >> Makes certain sense. >> >> Seems like the default value would have to be nil, though, if it's >> passed as an optional argument to a function. > > Yes. But given that this isn't a simple boolean, what it means when > disposition = nil is open to interpretation, and the doc string should > say clearly what that means. Sure. >> Are we okay with nil meaning "thread's buffer is killed and sent a >> signal", t meaning "thread's buffer is never killed" and 'silently' >> meaning "thread's buffer is killed without a signal"? > > Maybe. I'm not sure there's a definite agreement about the default. Here's a patch that implements the new default. I think there are no places in the manual where exceptions to killing a buffer are exhaustively described, so I just updated the description for 'make-thread'. LMK if something was missed.