Thus said Stephen Gildea on Wed, 20 Nov 2024 20:03:29 -0800: > I am building and testing with Ubuntu 24.04 on x86_64 hardware.
I'm building and testing on OpenBSD 7.5 amd64. Very strange that you get different results. What's even more strange is what the backtrace reveals---it would be interesting to inspect the value of nsc->ns_readfd. Are all the failures the same "bit out of range 0 - FD_SETSIZE", or are there others? What configure options do you have enabled? Perhaps inc on your system is going down a different code path than mine? nmh configuration ----------------- nmh version : 1.8+dev host os : x86_64-unknown-openbsd7.5 compiler : cc compiler flags : -g -O2 -Wall -Wextra linker flags : -Qunused-arguments preprocessor flags : source code location : . binary install path : /usr/local/bin libexec install path : /usr/local/libexec/nmh config files install path : /usr/local/etc/nmh man page install path : /usr/local/share/man docs install path : /usr/local/share/doc/nmh RPM build root : ./RPM backup prefix : , transport system : smtp spool default locking type : fcntl default smtp server : localhost SASL support : no TLS support : yes OAuth support : no It seems that if I want to reproduce this I'll have to wrangle up a Linux VM somewhere. At any rate, I'll take a look at your new patch when I get some free time. If it's a better approach it would make sense to take that instead of mine, even if I have been using mine successfully for a while now. I did apply your patch to the same commit from master and ran it through the same iterations of test-pop and there were no issues, but then, I have no issues with my patch either. :-) Andy
