- don't assume an implicit "listen on socket" when another listen was specified in a server block - allow servers to share the sockets they're listening on - fix unix_socket_name; now it's actually used - remove `listen on socket off' and `unix_socket off' options that are now no longer useful - unify the internal handling of the sockets, i.e. don't consider unix socket special. ok tracey@

h is free'd by add_addr in case of error, but we still leak the getaddrinfo result. while here make the check strictier: addr_addr() fails with -1.

these functions are expected to return -1 on failure, be strict about it.

forgot it set to a low value when testing the resizing, the left way too low. prodded by stsp. ok stsp

ok thomas_adam, op@

Add a delay after starting the server background process to keep server and client from racing against each other, which would lead to failures when the client ran before the server. ok op

ok op@

asctime_r() is not a source of valid date strings for email. Borrow code from OpenSMTPD instead for this purpose. Problem reported by naddy, who also found the smtpd code to use.

This should make it somewhat easier to keep track of discussion threads when someone replies to a commit notification to discuss the changes.

When multiple notification targets are configured only the first target received notification message content. The subsequent targets would read EOF when trying to read the content. Fix this issue and add regression test coverage for it, exercising existing test code with a gotd.conf file which contains two notification targets.

In particular, this prevents anonymous user accounts from discovering the existence of other private repositories served by gotd by correctly guessing the name of a private repository. They still wouldn't have read or write access but in some cases even knowledge about the existence of a particular repository could be cause for concern. ok op@

based on an initial patch by Matthias Schmidt

Specifically, when 'gut send' suggests 'fetch and rebase required', new users might try to use 'got fetch' directly followed by 'got rebase' without first updating the work tree to newly fetched commits. Got would then say "main is already based on main" without any hint for a way out. Hopefully, pointing users at the update -b command will make them search the manual for details.