[00:00] Noldorin: yeah, that doc is in the main branch - has been for a number of years :) [00:00] jelmer, does it though? i didn't think bzr init actually copies data from any other branch [00:00] ok [00:00] Noldorin: it copies data, but just the data that is relevant *for the new branch*. It doesn't copy a lot of other data that would still be valid because other branches, either in the same repository or in stacked branches, rely on it. [00:01] s/valid/necessary/ [00:01] ah [00:02] jelmer, eek, what's the command to do a shallow branch again? i'm just trying to get lp:bzr now, but i don't want it to be however many GB the full rev history would fill on my system heh [00:03] jelmer, do i just want a stacked branch maybe? [00:03] Noldorin: FWIW the repository is 70Mb, not GB [00:04] Noldorin: --stacked [00:04] oh right [00:04] i thought it would be a lot larger! [00:04] that's surprisingly small [00:04] jelmer, guess you do a lot of metadata compression... [00:23] Noldorin: branch only copies referenced history [00:23] Noldorin: that is very different to repository level gc [00:23] Noldorin: which has to delete content other concurrent processes may be looking at [00:24] Noldorin: someone writing gc can of course layer on all the selective-copy work that branch layers on too. [01:41] eek [01:41] okay [01:42] lifeless, has any of the implementation vaguely described in doc/develoeprs/gc.txt been started in fact? [01:52] Noldorin: I don't know, sorry - I haven't been following bzr dev closely for oh 3 years now [01:52] ah [01:52] Noldorin: it's not conceptually hard to do [01:52] Only thing I ever even heard rumors of was that old gc plugin, and if it's even available anywhere it was so long ago that it probably wouldn't fit without substantial rework anyway. [01:52] Noldorin: but it does involve working at core levels of data storage to ensure correctness around locking / visibility etc [01:53] lifeless, yeah, i wouldn't know where to begin with that... === wedgwood is now known as Guest58173 [02:08] lifeless, fullermd, probably infeasible for someone who doesn't know the internal structure of bzr still, no? [02:11] Structure is an illusion. === wedgwood is now known as Guest92845 [06:44] jelmer: it just thinks there were no changes [06:44] mathrick@megumi ~/elisp/dist/git/multiple-cursors $ bzr diff -c -1 [06:44] mathrick@megumi ~/elisp/dist/git/multiple-cursors $ === Guest92845 is now known as mew === mew is now known as wedgwood [15:47] hey fullermd [15:47] (if you're there) [16:41] lifeless, jelmer ? [16:44] Noldorin: it's a rle of irc, to just ask your question, rather than trying to handshake specific people [16:44] mgz, no. i need specific expertise [16:45] mgz, and i know who dealt with these things [16:45] the the mailinglist is a better venue. it's the middle of the night in nz for instance [16:45] *then the [16:47] ah yes [16:47] maybe [16:58] Noldorin: it's still a better idea to ask directly, even if you ping specific people, to bootstrap the talk, rather than going "you here?" "Yeah, what's up?" "Oh, I was away still around?" [17:02] actually, i do have a general question anyway: [17:02] what's the best way to learn about the internal structure of bzr? [17:02] the way it uses metadata [17:03] the formats utilised [17:03] etc. [17:03] Noldorin: for some things, there will be blueprints and design docs on the wiki and launchpad [17:04] ah [17:04] but I'm not familiar with any central place which would have complete, current and accurate picture [17:19] hmm [17:19] guess i'll just browse the wiki then [17:49] mathrick, any areas/pages in particular you might be able to point me to? :) [17:49] mathrick, i want to work on a gc/cleanup implementation after i familiarise myself with things [17:51] Noldorin: nope, sorry, I can't claim anything resembling enough familiarity to be able to tutor people [17:52] fair enough [17:58] Noldorin: use the source, that's the only up-to-date version. The tests though should give you examples for anything you may dream about ;) [20:17] vila: what I always find tricky in large systems with involved and tricky design like bzr is finding out why things are, and which parts are the important bits you need to undertsand, and which are just details [20:18] unfortunately source usually doesn't tell you that [20:32] Noldorin: the code [20:32] Noldorin: there are design docs in the source tree; and there are design docs in the classes [20:33] Noldorin: hi [23:25] lifeless, in the classes? [23:50] lifeless in the actual code you mean?