Re: [Monotone-devel] problem with update after merge (was: Re: db query error on propagate, monotone 0.16)

[Nathaniel Smith]

On Fri, Jan 21, 2005 at 09:06:36AM +0100, Georg-W. Koltermann wrote:
> It turned out to be more complicated than I thought.  The buglet does
> not appear if I only have one changeset in the branch that I'm merging,
> er, propagating.  It does appear with two, and probably with more.
> 
> I am attaching a script and the corresponding output exhibiting the
> problem.  The script is for 2d2a7e250d78513c9b1118616761bf96a529053d,
> earlier releases had a slightly different command syntax.

Oh!  I know what this bug is!  Huh, I thought we had a test for it
already, but I don't see one.  Added yours now; thanks.

The problem is that 'update' doesn't handle discontinuous branches.
What's happening is that 'propagate' is noticing that the head of
first-branch is a direct descendent of the head of main-branch.
Therefore, instead of trying to merge them, which would be silly, it
just certs the head of first-branch into main branch, as it says:
> monotone: no merge necessary; putting
> 75e624174b4db1eb1deba0776307f499d8db07ec in branch 'main-branch'

The result is that we have A -> B -> C, where A and C are in
main-branch but B is not.  One broken thing about the current update
algorithm is that it doesn't know how to jump past such a break in a
branch; this is will be fixed when we get around to rewriting the
update target selection algorithm.

As a workaround in the mean time, you can use 'monotone heads' to see
the real update target, and then 'monotone update <that revision>' to
update to it.

-- Nathaniel

-- 
"...these, like all words, have single, decontextualized meanings: everyone
knows what each of these words means, everyone knows what constitutes an
instance of each of their referents.  Language is fixed.  Meaning is
certain.  Santa Claus comes down the chimney at midnight on December 24."
  -- The Language War, Robin Lakoff