Jeff King <peff@xxxxxxxx> writes: > On Sat, Feb 23, 2008 at 01:56:38PM -0800, Junio C Hamano wrote: > >> > Well, the trap I wanted to avoid is that "/bin/sh" specifically is >> > broken. But yes, I think losing the ability to run the tests from the >> > commandline is bad. >> >> We never had it. "make t3404-rebase-interactive" is the way to >> do so. > > Well, that is a surprise to me, since I have been using it for almost > two years with no problems. And I know I am not alone because others > have posted snippets on the list invoking a test via "sh -x", including > both Dscho and you. > > Is there a good reason not to do so? I think our messages are crossing, but (1) currently we do not export much from t/Makefile and the only people who can get affected are on platforms that do need custom configuration, and the difference being subtle and rare makes it more surprising and harder to diagnose when the difference does matter, (2) I'd like to place some stuff in t/Makefile in such a way that no tests that runs a server that listens to a network port is not run by default, among other things, which means the difference between "sh tXXXX-name.sh" and "make tXXXX-name" will get bigger not smaller. - To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html