Le 15/07/2020 à 16:55, Junio C Hamano a écrit :
Jean Abou Samra <jean@xxxxxxxxxxxxx> writes:
$ git bisect run ./non-existent.sh
running ./non-existent.sh
/usr/lib/git-core/git-bisect: 247: ./non-existent.sh: not found
Bisecting: 2 revisions left to test after this (roughly 1 step)
[6fc732a2116e2c42b0431bb7e2a21719351af755] Fix typo in docs: 'created
by th' -> 'created by the' (GH-21384)
running ./non-existent.sh
/usr/lib/git-core/git-bisect: 247: ./non-existent.sh: not found
Bisecting: 0 revisions left to test after this (roughly 0 steps)
Yes, it would be nice if "git bisect run" can reliably tell that it
got a "not found" error and not a "test performed by the script did
not pass" and stop at the first failure.
On the other hand, the "./non-existent.sh" script could be part of
the tracked contents (i.e. some revisions have it and the working
tree has it when they get checked out, some revisions don't and the
working tree does not have it), and the user is trying to find the
first revision that stopped having a working script in its tree. In
such a case, the script that does not exist and the script that fails
need to be treated the same way by "git bisect run" as failures.
So... I dunno.
Perhapsa --not-found-as-failure option could help?
Best,
Jean Abou Samra
