known breakage: revision range computation with clock skew
authorJunio C Hamano <gitster@pobox.com>
Sun, 3 Feb 2008 07:47:22 +0000 (23:47 -0800)
committerJunio C Hamano <gitster@pobox.com>
Sun, 3 Feb 2008 08:25:52 +0000 (00:25 -0800)
commit991c3dc79f21a52209f9f41a52909cbb9462de1a
tree5d5b713af620dc427f3a432776fa097a16252c40
parent11d54b8b9a78d7236916883c09e165fdc3768f00
known breakage: revision range computation with clock skew

This is the absolute minimum (and reliable) reproduction recipe
to demonstrate that revision range in a history with clock skew
sometimes fails to mark UNINTERESTING commit in topologically
early parts of the history.

The history looks like this:

o---o---o---o
one         four

but one has the largest timestamp.  "git rev-list four..one"
fails to notice that "one" should not be emitted.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
t/t6009-rev-list-parent.sh [new file with mode: 0755]