On Tue, 2008-07-29 at 23:48 +0000, Jelmer Vernooij wrote:
> Trying to merge fails on bzr trying to access file id
>
> 951@1811c9d2-c306-0410-128--
> e57aa55c946:trunk:conduit%2Fmodules%2FFileModule%2FFileModule.py
>
> in revision
>
> john@nzjrs-uni-20080604054725-xb4vevcba6m9c8u6
>
> this revision id doesn't appear anywhere in branch "1"'s log and nowhere
> in the inventory xmls.
>
> It does exist in "2" though where it is the parent of a right hand side
> parent of a revision that is also in "1",
> <email address hidden>
>
> it seems to me like this is a case where merge doesn't fetch all the
> ghosts into the target repository before doing the actual merge.
Well, merge does a fetch(), not a fetch-ghosts, because ghost checking
is O(history) and thats a problem.
One possible fix, if it truely is just a ghost issue is to have a
error-handler do a JIT fetch for this sort of thing.
On Tue, 2008-07-29 at 23:48 +0000, Jelmer Vernooij wrote: c306-0410- 128-- trunk:conduit% 2Fmodules% 2FFileModule% 2FFileModule. py uni-20080604054 725-xb4vevcba6m 9c8u6
> Trying to merge fails on bzr trying to access file id
>
> 951@1811c9d2-
> e57aa55c946:
>
> in revision
>
> john@nzjrs-
>
> this revision id doesn't appear anywhere in branch "1"'s log and nowhere
> in the inventory xmls.
>
> It does exist in "2" though where it is the parent of a right hand side
> parent of a revision that is also in "1",
> <email address hidden>
>
> it seems to me like this is a case where merge doesn't fetch all the
> ghosts into the target repository before doing the actual merge.
Well, merge does a fetch(), not a fetch-ghosts, because ghost checking
is O(history) and thats a problem.
One possible fix, if it truely is just a ghost issue is to have a
error-handler do a JIT fetch for this sort of thing.
-Rob www.robertcolli ns.net/ keys.txt>.
--
GPG key available at: <http://