Bug: hg -R .hg/patches incoming/pull ../queue.hgq adds spurious heads and changesets

[Peter Arrenbrecht]

I have reported it:

  http://www.selenic.com/mercurial/bts/issue820

and a fix is nearly ready - it is blocked on:

  http://www.selenic.com/mercurial/bts/issue821

-peo

On Oct 30, 2007 5:13 PM, Peter Arrenbrecht <peter.arrenbrecht at gmail.com> wrote:
> Hi all
>
> When using "pull" or "incoming" instead of "unbundle" against a bundle
> of a managed .hg/patches, hg reports all of the base repo's changesets
> as changesets against the queue again (as a different head). When
> pulling, this wrecks the queue repo.
>
> In short:
>
>
> cd base
> hg -R .hg/patches bundle --base=null ../queue.hgq
> cd ../copy
> hg -R .hg/patches incoming ../queue.hgq
>
>
> results in
>
>
> comparing with ../queue.hgq
> 0       main: one added. - peter 1 second
> 1       queue: two.patch added. - peter 1 second
>
>
> where the first line is rubbish. (I have a compact default log format). Doing
>
>
> hg -R .hg/patches heads
>
>
> afterwards yields
>
>
> changeset:   1:4b1485a012be
> tag:         tip
> parent:      -1:000000000000
> user:        peter.arrenbrecht at gmail.com
> date:        Tue Oct 30 17:10:04 2007 +0100
> summary:     queue: two.patch added.
>
> changeset:   0:dcc6878641ad
> user:        peter.arrenbrecht at gmail.com
> date:        Tue Oct 30 17:10:03 2007 +0100
> summary:     main: one added.
>
>
> The attached bash script demoes the whole thing.
>
> Should I report this as a bug?
> -peo
>