Bibisecting: Over 1000 bibisects served!

[Bjoern Michaelsen]

Hi,

On Mon, Oct 19, 2015 at 10:41:19AM -0500, Norbert Thiebaud wrote:
> No it is not irrelevant. with the bad bibisect commit, say
> c49af8fc6406f9e7e8e0b1dcebee6df87bdeb9aa
> you immediately have the answer you seek:
> https://gerrit.libreoffice.org/gitweb?p=bibisect-macosx-64-5.0.git;a=commit;h=c49af8fc6406f9e7e8e0b1dcebee6df87bdeb9aa

I guess Stephans point is that if somebody just throws a "source
sha:6586da0631ddcfd704538b0e1cf96d2ea0be7cd9" line about there is no way of
knowing what that means without doing research into what bibisect repo it was
from and how the bibisect log looks like.

Your "lets write all commit hashes in the bibisect commit" improvement is nice
for devs bisecting themselves and then continuing to fix the stuff. But once
there is a handoff between a bibisecter and a bug fixer, a source commit range
in the form 

 "<source-hash-of-last-known-good>..<source-hash-of-first-known-bad"

is the most universal, short and unambigous format to communitcate that
information.

Best,

Bjoern