Subject: Re: lp:~bzr/bzr/current - msg#01770

Previous Message by Date: click to view message preview

Re: Trying out nested trees

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Jelmer Vernooij wrote: > I thought I'd give the current nested trees support a spin Thanks for taking an interest. > Overall, I really like how easy it is to work with subtrees; > it's much more intuitive than svn externals or git submodules IMSNHO. That's great! > Adding a subtree without changing that subtree itself gives the > following output during commit: > > Committing to: > /home/jelmer/tmp/subtree/ > Committing to: > /home/jelmer/tmp/subtree/specimen/ > aborting commit write group: PointlessCommit(No changes to commit) > added specimen > Committed revision 1. I really wish this write group abort reporting has not been added. > The PointlessCommit error for the subtree is a bit confusing. It used to be swallowed silently, and I think it still should. Preventing it would require look-before-you-leap. > "bzr ls" also seems to not handle tree-references inventory entries yet > (bzr: ERROR: unknown kind 'tree-reference') I am adding support on a command-by-command basis: http://bazaar-vcs.org/NestedTreeProgress ls was not planned, but should be easy to support, once composite trees land. > Push seems to error out somehow trying to push the branch with the subtree: > > ganieda:~/tmp/subtree% bzr.dev push ../subtree2 > bzr: ERROR: Not a branch: "/home/jelmer/tmp/subtree2/specimen/". Push is on the TODO list. > "bzr add <existing-branch>" is a noop, e.g. if foo is a branch: This is the standard behavior of add. If foo is a branch, we try to add files *to* foo. > $ bzr add foo > $ bzr st > unknown: > foo > > I think this should probably warn the user that they want to use "bzr > join --reference". I don't think we want to be recommending nested trees until they're out of beta, at least. --reference is a hidden option for that reason. Aaron -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkn4o+EACgkQ0F+nu1YWqI20+wCdFF0WmdUnwOVZ3Pptfi5lnrAM z5wAn1RL17i5lZ+qupyj48z2diy08xBf =SdMT -----END PGP SIGNATURE-----

Next Message by Date: click to view message preview

Re: Better name for dpush wanted

Jelmer Vernooij writes: > bzr core provides "only" 61 visible commands of those 122. Sure, but people are going to want plugins. rebase, gtk, and bzrtools are all pretty popular, and launchpad will become more so over time. > I agree that having a lot of commands can be overwhelming to > users. I don't see a lot of things that can be removed or would be > good candidates for merging though (bzr --no-plugins help > commands). * Dramatic reduction possible for sure: conflicts, deleted, ignored, ls, missing, renames -> ls alias, nick, plugins, version, whoami -> config clone -> <delete> bind, checkout, pull, switch, unbind, update -> checkout tag, tags -> tag[1] init, init-repository -> init[2] pull, merge, remerge -> merge[3] uncommit, revert -> revert info, revno, root, status -> status[4] * There's gotta be some consolidation possible here: break-lock, check, clean-tree, ignore, pack, reconcile, reconfigure, remove-tree, split, upgrade, version-info, view -> manage, info I think you should fairly easily be able to reduce the number of core commands to 40, with some effort, 35. Footnotes: [1] It's actually harmless to have both the singular and plural form with the obvious semantics, but it's the principle of the thing. [2] Cf. my immediately previous post. [3] "pull" is a horrible name for "mirror" semantics. Mostly it belongs in checkout/update, some in merge. [4] "status" is not a great name for the information provided by info, but it seems reasonably apt as a switch: "bzr status --branch", for example, defaulting to --files.

Previous Message by Thread: click to view message preview

Re: lp:~bzr/bzr/current

2009/4/29 Bob Tanner <tanner@xxxxxxxxxxxxx>: > When I tried to update lp:~bzr/bzr/current I get the follow error: > > bzr: ERROR: Cannot lock LockDir(lp-<long > numbers>:///~bzr/bzr/current/.bzr/branchlock): Transport operation not > possible: readonly transport > > Can someone tell me what command I should be using to update current? > > Please confirm I still have write access to current? This sounds like it may not be using the right ssh transport? Can you use an explicit url like bzr+ssh://bobwhatever@xxxxxxxxxxxxxxxxxxxx/~bzr/bzr/current, or can you run bzr launchpad-login? -- Martin <http://launchpad.net/~mbp/>

Next Message by Thread: click to view message preview

Re: lp:~bzr/bzr/current

On 29/04/2009 21:13, Bob Tanner wrote: After launchpad-login all is working. Just seems a weird error for not being logged in. I agree, I saw this too. And internal error with stack trace dump if LP isn't available (ie. on time out of connection). Perhaps these error cases (quite common) could be handled in a more user friendly way... :-P -- Philippe Lhoste -- (near) Paris -- France -- http://Phi.Lho.free.fr -- -- -- -- -- -- -- -- -- -- -- -- -- --