On Mon, Nov 04, 2019 at 12:19:33AM +0100, Darshit Shah wrote: > Hi Francesco, > > This issue has indeed been reported multiple times in the past. While it is > confusing, the text "Success" is out of our hands. It is the value of > `perror()` after we encounter that a system call resulted in the error.
The value of errno does not live long. After any library call it can be destroyed. So the code in http.c case FWRITEERR: case FOPENERR: /* Another fatal error. */ logputs (LOG_VERBOSE, "\n"); logprintf (LOG_NOTQUIET, _("Cannot write to %s (%s).\n"), quote (hstat.local_file), strerror (errno)); seems broken. The value of errno must be saved if its value is needed later.