Eli Zaretskii writes: >> Cc: 74642@debbugs.gnu.org >> Date: Wed, 04 Dec 2024 15:41:18 +0200 >> From: Eli Zaretskii >> >> > From: Diancheng Wang >> > Cc: 74642@debbugs.gnu.org >> > Date: Wed, 04 Dec 2024 16:22:51 +0800 >> > >> > > I guess you mean "C-x k RET"? >> > >> > Yes, I mean "C-x k RET" >> > >> > > >> > > When I do that, Emacs asks me whether to kill the process associated >> > > with the buffer, and if I say YES, the indicator does disappear >> > >> > Yes, the steps are same with me. >> > >> > > (sometimes after a second, which I guess is needed for Emacs to kill >> > > the process). Isn't that what you see? >> > >> > No, the indicator does not disappear after I killed "*gud-myprogram*" buffer. >> > By the way I started emacs using "emacs -Q". >> >> So did I. >> >> > The attachment is my running emacs information (the buffer content >> > created by report-emacs-bug) >> >> Looks like some kind of regression in Emacs 31. It works in Emacs 30. >> If you could bisect to find the offending commit, it will be >> appreciated. > > Crystal ball says that the regression was caused by the change you > submitted (and I installed) in bug#73687. It added a call to > gud-hide-current-line-indicator to gdb-starting and to gud-sentinel. > Could you try reverting those changes and see if that solves the > problem in Emacs 31? Yes, There is a regression in my patch: the arrow on the source line is not stopped displaying when "*gud-myprogram" got killed. But Emacs 30 and Emacs 31 still have another issue: gud-highlight-current-line-overlay is not destroyed when "*gud-myprogram*" got killed. To reproduce it, you need to evaluate (setq gud-highlight-current-line t). I updated my patch to fix this 2 issues, see attachment.