This cuts 30% off the running time of the test suite: I had 5:43.58 before and 4:00.52 afterwards. So, not bad. You could argue that this should only be enabled when it's not a release build. Thoughts on that?
I think this needs some #ifdefs, unless you've verified that it can be cross-compiled to Windows (I recommend mingw-w64, as seen in Debian unstable) and/or looked it up on MSDN. Either throw in some configure checks for setsockopt, IPPROTO_TCP and TCP_NODELAY, or approximate it by #ifdef G_OS_UNIX? Other than that it seems good. We like fast tests. (In reply to comment #0) > You could argue that this should only be enabled when it's not a release build. I'd rather not have release builds test themselves differently. Perhaps you could make it controllable by an environment variable or its own --enable thing, and have Gabble use the default (i.e. not NODELAY) during distcheck?
(In reply to comment #1) > Either throw in some configure checks for setsockopt, IPPROTO_TCP and > TCP_NODELAY, or approximate it by #ifdef G_OS_UNIX? I did this. :-) > (In reply to comment #0) > > You could argue that this should only be enabled when it's not a release build. > > I'd rather not have release builds test themselves differently. Perhaps you > could make it controllable by an environment variable or its own --enable > thing, and have Gabble use the default (i.e. not NODELAY) during distcheck? Okay I made the default be not having NODELAY, but setting GABBLE_NODELAY will enable it for a test. I even wrote about it in tests/README.
Looks great. We like faster tests.
Merged, thanks.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.