At 14:49 +0100 07 Nov 2020, Matthias Apitz <g...@unixarea.de> wrote:
Can you please elaborate a bit what's not a good idea with this patch? It deals exactly with returning waitpid() -1 and errno=ECHILD, i.e. with the exact problem we have here, why ever the child was and/or SIG_CHLD was lost. Thanks
Mutt isn't just trying to determine *that* the child exited, but also *how* it exited, so that the user can be informed if something apparently went wrong. In the case of getting an ECHILD error we only know that the child exited and that the status was somehow lost.
In your case (at least the times that you've examined) it seems that the child exited successfully, but if that status can be lost I can almost guarantee that in some cases the lost status will be an unsuccessful one. That the exit status is getting lost is certainly a problem, but just ignoring the problem and hoping that the child process finished successfully isn't a real solution.
signature.asc
Description: PGP signature
