[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Monotone-devel] merge weirdness...
From: |
Christof Petig |
Subject: |
Re: [Monotone-devel] merge weirdness... |
Date: |
Wed, 27 Apr 2005 09:11:31 +0200 |
User-agent: |
Mozilla Thunderbird 1.0.2 (X11/20050404) |
Richard Levitte - VMS Whacker schrieb:
> I've done quite a few merges and propagates from net.venge.monotone to
> net.venge.monotone.command-specific-options in the last few days, and
> found it a bit weird that the propagate (and lately, the merge) seemed
> to try to reapply exactly the same changes from one time to the other.
> Wow! With such an old ancestor, I suddenly understand why monotone
> keeps on running into the same conflicts time and time again.
>
> For some reason, I would say that the last propagate should have
> chosen f188f1efa53fc5e1f7aa1313f5d5af8577a0dac9 for a common ancestor.
>
> Maybe I've misunderstood what it means to have common ancestor. I'm
> just pretty sure that it would have been just as correct, and would
> have had much fewer conflicts or repeated changes.
That's exactly the reason why I always (!) use the four argument
explicit_merge when propagating (monotone-viz is really nice to pick the
correct nodes). Njs told me that the lca[d] algorithm is mathematically
correct but does not always produce the human desirable result. It
sounded a bit like a TODO:
http://lists.gnu.org/archive/html/monotone-devel/2005-01/msg00131.html
Since then I got the impression that everybody else has no problems with
the current algorithm but I myself found it unusable in 90% of the times
I tried.
Christof
signature.asc
Description: OpenPGP digital signature