More merge base discussion [was: monotone's LCA+DOM algo for selecting a merge base]

[John A Meinel]

Aaron Bentley wrote:
> John A Meinel wrote:
>>> Because ultimately a
>>> human who understands what the different revisions are, can probably
>>> pick the best base.
> 
> I wish I could agree, but whenever we get a weird merge and I dive into
> graph-ancestry to figure out whether the base is wrong, I have a hell of
> a time figuring out whether the auto-selected base is right or wrong.
> 
> Aaron

But you are trying to figure it out based on applying the logic you
thought you were applying. (Like trying to say, if I were bzr what base
would I pick).

I'm saying that a human would pick a particular base either because the
topology looked good enough, etc.

I do think we need to get our graphs looking a little bit clearer before
a human could understand them, though. Dot doesn't do a great job when
you have 3000 revisions. It might do okay if you have 30 or 300. And I'm
curious what would happen if we used 'bzr viz' style, but had it use the
first column as THIS, and the second column as OTHER. I think the
histories might end up looking quite nice.

John
=:->

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 249 bytes
Desc: OpenPGP digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20060220/9e9a6005/attachment.pgp