[07:51] <mlischke> Hi, I have a bzr repository on a server without working tree and when I run "bzr status" (or "bzr merge.." what is my original intent) I get the error "bzr: ERROR: No WorkingTree exists for..".
[07:51] <mlischke> Shouldn't that work also for tree-less branches?
[07:52] <lifeless> no
[07:52] <lifeless> status reports on the tree status
[07:52] <lifeless> it has no meaning for tree-less branches
[07:52] <lifeless> bzr info will work
[07:53] <mlischke> hmm, forget bzr status then, it was just a test, actually I want to merge another branch into this one
[07:53] <mlischke> (bzr info works as expected)
[07:53] <lifeless> well you'll need a tree :)
[07:58] <mlischke> sounds a bit … weird, I can create a branch without working tree, shoudn't I be able to merge from the parent tree to the copy also without a tree?
[07:59] <lifeless> no, because merge's job is to prepare an updated tree.
[07:59] <lifeless> when you commit the merge you can push the commit to a treeless branch
[08:02] <mlischke> Ok, so what can I do to to make the copy take all changes from another branch? Create a tree in the copy, merge, remove tree?
[08:04] <lifeless> mlischke: either do the merge elsewhere and push the result of the merge to the branch, or checkout the branch and do the merge in the checkout.
[08:05] <mlischke> hmm, actually, this was my first approach, I have a checkout of both branches, but that gives me a criss-cross
[08:05] <lifeless> mlischke: for instance; if you want to do this directly on the server; bzr checkout --lightweight $(branchpath) /tmp/foo; cd /tmp/foo; bzr merge $(otherbranch); bzr commit -m "Merged otherbranch";cd ..; rm -rf foo.
[08:06] <mlischke> ok, let me try this on the server...
[08:06] <mlischke> Previously I only tried locally
[08:06] <lifeless> mlischke: criss-cross merges are a result of having an ambiguous history, not of the trees
[08:07] <lifeless> you'll get that everywhere until it's resolved.
[08:07] <mlischke> hmm, so it wouldn't help to merge on the server?
[08:07] <mlischke> I always have to resolve criss-cross issues?
[08:08] <mlischke> My hope was it would be simpler to solve when doing it on the server, but if that wouldn't help I can as well go through the hassle and clean it up locally
[08:14] <mlischke> lifeless: thanks, I see the same criss-cross merge problem on the server...
[08:14] <lifeless> mlischke: http://doc.bazaar.canonical.com/bzr.dev/en/user-reference/criss-cross-help.html
[08:15] <mlischke> yes, thanks, have read this already
[08:15] <mlischke> and use weave, but got +50 conflicts, so I was looking for an easier approach
[08:15] <lifeless> so yes, you will see it everywhere until its been resolved (by merging into one or the other side)
[08:15] <mlischke> seems I have to bite the bullet
[08:15] <lifeless> sorry yeah
[08:16] <lifeless> this tends to happen if you merge branch A to branch B and B to A without picking up the prior merge of A to B
[08:17] <mlischke> It was a chain like this: A -> B, A -> C, now C -> B (as preparation for a later merge back to C)
[08:18] <lifeless> that can do it
[08:18] <lifeless> one way you might be able to fix it
[08:18] <mlischke> C -> B is rather C merge to B, not a branch
[08:18] <lifeless> merge C->B with -r - and choose the revision of C right before the merge of A->C
[08:18] <lifeless> commit that
[08:19] <lifeless> then merge C->B without -r
[08:21] <mlischke> thanks, will try that
[11:34] <icebrain> Hi! I'm thinking of setting up a readonly bzr server. Is it safe to give access using bzr_ssh_path_limiter?
[11:35] <icebrain> safe enough to, say, publish the access information publicly?
[12:25] <mgz> should be, you can just expose over http, which is reasonable performance for smaller repos
[17:03] <wedgwood> anyone know of good examples of how to do the bzrlib equivalents of status (minimally checking for changes), revert, and clean-tree?
[17:11] <beuno> wedgwood, I can help after lunch
[17:11] <wedgwood> beuno: much appreciated
[17:12] <wedgwood> I was just about to grab the bazaar source, but I'm nearly certain there's a more concise way
[17:12] <beuno> wedgwood, have you seen this?  http://wiki.bazaar.canonical.com/Integrating_with_Bazaar
[17:12] <beuno> it won't have exactly what you're looking for, but it may tp you in the right direction
[17:12] <beuno> IIRC, what you want to operate on is the working tree
[17:14] <wedgwood> no, I hadn't seen that. It does cover the status/diff bits. I'll see if I can find the rest.
[17:14]  * beuno goes eat
[17:15] <wedgwood> I see WorkingTree.revert (undocumented). Looks promising.
[18:01] <wedgwood> beuno: I found everything I need. thanks for the pointer
[18:19] <beuno> wedgwood, w00t, np