[04:56] <mark06> why the same branch is 7MB fresh new from Launchpad and 30MB in the copy where I've recommitted whole branch?
[04:57] <mark06> I thought uncommit mean it
[04:58] <spiv> Uncommit doesn't delete the commit from the repository
[04:58] <spiv> It just changes the branch to no longer refer to it.
[04:58] <bob2> did 'bzr gc' ever get implemented?
[04:58] <mark06> annoying
[04:59] <mark06> how to really uncommit it? or delete from repository if you prefer
[04:59] <spiv> Also, you will probably find 'bzr pack --clean-obsolete-packs' reduces the size of the bzr directory (note that bzr will do that housekeeping automatically from time to time without you needing to do it manually)
[05:00] <mark06> does it delete uncommitted commits?
[05:00] <spiv> Until someone implements the hypothetical 'bzr gc' command bob2 mentions, you have to do it manually: create a new repo, and branch into that (omitting any branches and thus revisions that you don't want).
[05:00] <bob2> no
[05:00] <bob2> you can reclone if you really care
[05:01] <lifeless> didn't someone implement gc in a plugin ?
[05:01] <spiv> (If you only have one branch, then this is simply 'bzr branch original-dir new-dir')
[05:01] <spiv> (and are using standalone branches, I should add)
[05:02] <mark06> well I only create branches with bzr init, does that qualify?
[05:02] <spiv> If you never used bzr init-repo, sure.
[05:04] <mark06> so it's basically bzr branch $each $each.new && rm -r $each?
[05:05] <mark06> right I never used bzr init-repo
[05:08] <mark06> btw, I was thinking of gathering together some scripts of mine which currently have their own branches, but if I use a single branch I'll lose individualized history for each script
[05:08] <spiv> Right.
[05:08] <spiv> There are ways to merge unrelated branches without losing history.
[05:09] <mark06> so I wondered if I could create some sort of "multi-branch", is that what repos are about?
[05:09] <spiv> Search the mailing list archives, IIRC the 'bzr merge-into' plugin may help
[05:10] <spiv> (Also "bzr merge -r0..-1 $unrelated_other" for simple cases)
[05:11] <spiv> No, shared repos are simply about efficiency.
[05:12] <spiv> (If you have N branches of a project, the vast majority of history will be exist in more than 1 branch, but if they share a repo each revision is only stored once.  This unsurprisingly saves a heap of disk space and makes things faster.)
[05:12] <spiv> Ugh, pardon the mangled grammar :)
[05:12] <fullermd> I think jam wrote a gc plugin.  A very long time ago.  Like pre-2a ago, if not earlier...
[05:13] <mark06> ok, but your suggestion will only work for past commits not new ones, right
[05:13] <fullermd> You can merge multiple unrelated branches together; I do it all the time.  Can't work with them independently after that though.
[05:14] <fullermd> (well, you can, on the original unjoined bits, but post-join work is all somewhat irrevocably joined)
[05:15] <mark06> well, actually I can bzr [q]log filename, then I'll see individual histories separately
[05:16] <fullermd> Yes, but I mean you can't turn that one joined branch back into N separate branches.
[05:18] <mark06> ok I mean about the whole idea of multi-branches, not needed at all... the only inconvenience is need to specify component in commit messages for global log not being confusing, but it's ok
[05:20] <mark06> either way, I think you can hack into uncommit and revert to unmerge separate branches... original commit dates could be kept with --commit-time in the new branch...
[05:21] <mark06> which I just found out after recommitting all my problematic branches :(
[05:21] <fullermd> Only by backing up to before they're merged together (which kinda beats the dolphin).
[05:23] <mark06> btw I have recommitted some branches due to a bug between MSVCRT and MSYS, but in the meanwhile I found out what seems a bug with qlog...
[05:24] <mark06> commit times were being kept 1 hour forward (timezone -2) but even after fixing that (timezone -3) qlog seems to still be confused...
[05:25] <mark06> maybe still referencing replaced  commits? I haven't found a pattern to reproduce and file a bug though
=== mark06 is now known as mark06-away
=== mark06-away is now known as mark06
=== mmrazik is now known as mmrazik|afk
=== mmrazik|afk is now known as mmrazik
[08:13]  * fullermd waves at vila.
[08:18] <jelmer> mgz!
[08:18] <mgz> morning
[08:20] <mgz> how are you today jelmer?
[08:23] <jelmer> in need of coffee
[08:23] <jelmer> mgz: how are you?
[08:24] <mgz> not quite arrived at coffee yet :)
[08:26] <vila> fullermd: \o
[08:26] <vila> hi jelmer, mgz ;)
[08:26] <mgz> hey vila!
=== lifeless_ is now known as lifeless
[08:40] <vila> lifeless: happy EOJ ;) Have fun !
[08:47] <lifeless> vila: thanks
=== mmrazik is now known as mmrazik|lunch
=== mmrazik|lunch is now known as mmrazik
=== maxb_ is now known as Guest72551
=== carif_ is now known as carif
[16:02] <Limit_> hi!
[16:02] <Limit_> I am unable to connect to launchpad using bzr
[16:03] <Limit_> It shows connection reset and I wanted to know if I could ask bzr to make requests at specific ports. Any help?
[16:24] <Limit_> SSH in my college is blocked. Can anyone suggest an alternative way to connect to Launchpad using bzr ?
[16:29] <vila> Limit_: then no write access for you :-/ http should be fine to read though
[17:10] <Limit_> vila: so I would atleast be able to fetch code??
[17:12] <mgz> Limit_: yup, you need to remove your launchpad login from your config and ignore bzr telling you you've not logged into launchpad
[17:13] <mgz> that means removing 'launchpad_username' from bazaar.conf and deleting the launchpad.net section of authentication.conf
[17:13] <Limit_> mgz: oh! that would atleast partly solve my problem! thanks :)
[17:14] <mgz> running `bzr version` will tell you where to find those
[17:14] <mgz> then when you branch lp:PROJECT it'll use the http link (which your college probably don't filter)
[17:15] <Limit_> mgz: new doubt. When  I try to upload code the next time, I'll have to add the launchpad_username and add info in authentication.conf, right?
[17:20] <mgz> Limit_: just `bzr launchpad-login YOURLPNAME` would be enough, but if this is a laptop and you're going on and off networks with ssh support that would still get annoying
[17:21] <mgz> what I had for a long time, was a pristine copy of trunk that was linked to the http location of the branch on launchpad
[17:22] <mgz> but then when I did feature branches that needed pushing rather than just pull, had my lp name in so push lp: used ssh
[17:24] <Limit_> mgz: yes i have a laptop and that was the reason I asked you :)
[17:25] <vila> interesting use case, worth a bug, this screams: configure me !
[17:25] <Limit_> vila: I don't understand what configuration changes would be useful in this case
[17:27] <vila> a way to switch from places with ssh and the others
[17:27] <vila> *ssh access
[17:28] <vila> did you try asking your admins to allow ssh access by the way ?
[17:29] <Limit_> Yes I did
[17:29] <Limit_> They simply denied.
[17:30] <Limit_> vila: about the configuration, we just need to write a script that swaps 2 configurations :) I have already created one for home and one for college. The script would name the required one as bazaar.conf or authentication.conf :)
[17:30] <vila> they denied blocking ssh ?
[17:31] <Limit_> no, they denied unblocking SSH
[17:31] <vila> oh, as in they refused (sorry, not a native) ?
[17:31] <Limit_> vila: Yes. we are not having SSH connection which stops me from pulling/pushing code.
[17:32] <vila> Limit_: otherwise, you're right, a simple script can do. You'll have to remember modifying the two copies in sync though
[17:32] <Limit_> vila: yes
[17:33] <fullermd> Well, the usual solution to such blocking is "just tunnel it over ssh"...   ;>
[17:34] <Limit_> fullermd: I could tunnel SSH over HTTP but, I don't think launchpad would accept the code that is pushed over HTTP
[17:35] <fullermd> Oh, well, you could just encode those as DNS TXT packets for the transfer.
[17:38] <Limit_> fullermd: I am sorry I didn't get you. Can you explain DNS TXT?
[17:38]  * fullermd is being a little facetious  ;p
[17:39] <fullermd> Isn't there a way you can make arbitrary location aliases on a branch?  There's also the bookmarks plugin.
[17:40] <fullermd> (you'd still have to spec the location when you're on whichever network you consider the special case, but it'd save you typing out the full deals)
[17:42] <mgz> just having the http link as the parent worked suprisingly well for me
[17:43] <mgz> wasn't often I wanted to do more than pull trunk and hack locally
[17:43] <Limit_> mgz: yes it will let me work on my code. It's just that for the time I am in college I'll have to submit patches.
=== deryck is now known as deryck[lunch]
=== deryck[lunch] is now known as deryck
[21:31] <janos__> in bzr explorer, is there a way to remove the [Toolbox] somehow?
[21:32] <janos__> i mean permanently, with configuration or something
[21:33] <janos__> the thing is, every time i click refresh it comes back
[21:34] <janos__> maybe that's not clear enough
[21:34] <janos__> here's what i mean:
[21:35] <janos__> create a branch: bzr init tmpbranch
[21:35] <janos__> open bzr explorer on it: bzr explorer tmpbranch
[21:35] <janos__> the top-right part is the "Toolbox"
[21:35] <janos__> it's something you hardly ever use (me, never)
[21:36] <janos__> you can click the [x] button, it will go away, but only until you hit refresh
[21:39] <janos__> a workaround that helps a bit: snap the Toolbox out of the frame, and then snap it back in -> this creates some sort of tabbed window, with the Toolbox and the Working Tree as tabs. This is better, because I can just switch to the Working Tree tab, the Toolbox will not be visible that way, and this layout is preserved even if I refresh
[21:39] <janos__> but i have to do this every time i launch bzr explorer
[21:45] <w7z> janos__: don't know off hand, but one way or another should be possible to remove it
[21:49] <janos__> w7z: so there is hope!
[21:50] <w7z> if nothing else, you can probably comment it out in the python
=== yofel_ is now known as yofel
[21:52] <janos__> w7z: probably
[21:54] <janos__> w7z: i need this for some non-programmer friends, it won't look very good if i have to tell them to go into the python code and comment out line X
[21:55] <janos__> well if there is no other way it's better than nothing
[21:56] <w7z> janos__: try line 205 in explorer.py
[21:57] <w7z> http://bazaar.launchpad.net/~bzr-explorer-dev/bzr-explorer/trunk/view/head:/lib/explorer.py#L205
[21:57] <w7z> comment out that in your local copy
[21:57] <w7z> or poke this function:
[21:57] <w7z> http://bazaar.launchpad.net/~bzr-explorer-dev/bzr-explorer/trunk/view/head:/lib/explorer.py#L708
[21:58] <w7z> that looks like it has preferences you can set
[21:58] <janos__> ok hang on
[21:58] <w7z> so try setting those preferences to false
[22:00] <janos__> thanks w7z, this is gooood!
[22:01] <w7z> you found the preferences and that did the right thing?
[22:02] <janos__> L205 just made the toolbox displayed popped out, the other one around L708 does what I want, I just set tbox_applicable = False ignoring all the ifs
[22:06] <janos__> w7z: i'm having the impression that i could set preferences somewhere, that way my friends don't have to edit python code
[22:07] <janos__> w7z: any ideas where that might be? maybe somewhere under ~/.bazaar ?
[22:15] <janos__> uhm, it's the DEFAULT_PREFERENCES['toolbox-on-status-view'] what i want to override to False, wonder if the setting is exposed somewhere in the UI or config file
[22:24] <janos__> w7z: got it! it's in ~/.bazaar/explorer/explorer.conf
[22:25] <janos__> i can set there the toolbox-on-* to false
[22:25] <janos__> thanks a lot w7z this is great!
[22:26] <janos__> gotta go now, bye all