| [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 | |
| === r0bby is now known as robbyoconnor | |
| [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 | |
| === stub` is now known as stub | |
| [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 | |
| === vednis is now known as mars | |
| === iBasic is now known as BasicOSX | |