Torsten Bögershausen <tboegi@xxxxxx> writes: > On 11.05.13 22:09, Junio C Hamano wrote: >> Torsten Bögershausen <tboegi@xxxxxx> writes: >> >>> I did, >>> the interesting thing is that the test passes with and without your patch. >>> (After enabling GIT_TEST_LONG and GIT_TEST_HTTPD in both cases) >> >> Strange. Do you see differences between the produced packed-refs >> file? > > The original version seems to look like this: > :1 666527db455708922859283c673094002092910b > :2 1e2acf73c6db881cfb1d56d67662e3d9260be2cf > [snip] > > The "fixed POSIX version" follows that style: > 666527db455708922859283c673094002092910b refs/tags/blablablablablablablablablablablablablablablablablablablablablablablablablablablablablabla-1 > 1e2acf73c6db881cfb1d56d67662e3d9260be2cf refs/tags/blablablablablablablablablablablablablablablablablablablablablablablablablablablablablabla-2 > [snip] That means whatever the test that comes after that set-up step to populate the packed-refs file does not need the packed-refs file. In fact, the test only sees if "clone" succeeds, without checking what refs are present in the resulting repository. The original repository lacks these 50000 tags due to the non-portable sed construct (it has a corrupt packed-refs file) and it may or may not have other refs, but because cloning an empty repository does not error out these days, the test just passes. Sloppy. -- 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