[Libreoffice-qa] Keyword "bisected"

[Bjoern Michaelsen]

Hi,

On Mon, Jan 05, 2015 at 08:23:20AM -0800, Joel Madero wrote:
> A couple people have told me they use the keyword to indicate that the
> exact (or nearly the exact) commit has been identified. 

I use it like that.

> If this is true...it's good to have (and would be another potential easy hack
> to move bibisected to bisected), but we need to share this information with
> developers so that it's actually useful information. Else, if developers are
> just querying for "bibisected" in whiteboard, they are going to miss bisected
> in keyword....

FWIW, I used it only on bibisected bugs, so at least where I set it, the bug is
"whiteboard: bibisected keyword: bisected" in the end, so that bisected bugs
are a subset of the bibisected bugs.

This was done when I did the other swipe over regressions etc. and was
discussed on the ESC:

 http://nabble.documentfoundation.org/minutes-of-ESC-call-td4127344.html

Back then, I looked for all the bibisected Writer regressions and identified
those that have a suspected single commit identified, because I wanted to have
an idea of how many bugs are triaged down to one commit. For Writer, it was
quite a huge set of the open bibisected regressions. This is a Good Thing as
shows that bibisecting works. This is a Bad Thing, as we likely should be
faster/more vicious in evaluating if the commit that caused the regression was
worth it (because it e.g. fixed a bigger problem). In some cases, it might make
sense to close the bug as WONTFIX, if there is no proper solution, and
reverting the guilty commit would bring back bigger issues.

Best,

Bjoern