Branch lifecycle statistics?

[Vincent Ladeuil]

>>>>> "Jonathan" == Jonathan Lange <jml at mumak.net> writes:

    Jonathan> I *am* interested in that information. But sadly, Bazaar
    Jonathan> does not provide any way to get at that information that I
    Jonathan> know of.

Hmm, well, Bazaar history recording is discrete, based on commits, not
on working trees creation so you can only use commit dates.

To estimate the lifecycle of a merged branch, based on say revno S and
merged into revno M, it may be tempting to consider that the branch was
created before revno S+1 (on the trunk), but that will be totally wrong
for people practicing daggy branching.

So you could approximate and start with the date of the first commit
(which will shorten the duration). But since that is for statistical
purposes and that the revno M date will often augment that
duration... you may want to try with that anyway and see if the skew is
really a problem (in which case, as for all statistical study, any user
of your data will be warned by your explanation :-)

   Vincent