[16:19] <ion_> Another tech talk about git: http://youtube.com/watch?v=8dhZ9BXQgc4 (http://chi-v14.chi.youtube.com/get_video?video_id=8dhZ9BXQgc4)
[18:03] <Keybuk> ion_: somewhat off-topic for here? :)
[18:04] <ion_> Yes. :-P
[18:06] <Keybuk> git is nasty
[18:07] <ion_> I’ve fallen in love with it. :-)
[18:07] <Keybuk> really, why?
[18:08] <Keybuk> I've yet to see anything it can do that other distributed vcs's can, except make me feel like I'm back in the dark ages of revision control (the Arch days)
[18:09] <ion_> It makes branches soooo nicely. I also like how you can ‘add’ files to the next commit while developing something and then finally do the commit. In the others, you’d have to collect the list of files to be committed while running the actual commit.
[18:10] <Keybuk> makes branches nicely?
[18:10] <Keybuk> nicer than "cp -a" ?
[18:10] <ion_> Hugely nicer than having each branch in a separate directory, adequately nicer than ‘bzr switch’.
[18:10] <Keybuk> it's identical to "bzr switch" isn't it?
[18:11] <Keybuk> (assuming the branches share the same repository)
[18:11] <ion_> Basically, but with a nicer UI.
[18:11] <Keybuk> IME, the git UI is the worst thing about it
[18:11] <Keybuk> like why would you ever want to make changes to something and not have them committed?
[18:12] <Keybuk> which is the default behaviour of git
[18:12] <ion_> Debugging lines
[18:12] <Keybuk> so by default, most of the code you write is debugging lines
[18:12] <Keybuk> and you want to only commit rarely?
[18:12] <ion_> I just like how it works more. Not everyone might.
[18:12] <Keybuk> that's a very kernel-optimised use case <g>
[18:12] <ion_> git commit -a isn’t very hard to write when you want what e.g. ‘bzr commit’ does.
[18:13] <Keybuk> I find the bzr model of "shelve the debugging lines while commit" much nicer
[18:13] <ion_> I also find it subjectively faster than bzr, but i haven’t done any actual benchmarking.
[18:14] <Keybuk> that was certainly true of network operations for older bzr versions
[18:14] <Keybuk> I don't think it's been true for a while
[18:14] <Keybuk> my main problem with git is it's optimised to assume that most of the operations you do are merges
[18:14] <Keybuk> and that you rarely commit any changes you make to the source tree
[18:14] <Keybuk> which is totally the opposite of any normal code tree
[18:15] <Keybuk> almost all changes I make are to be committed, and I rarely get any to merge :)
[18:15] <Keybuk> so git makes it harder for me to work with it
[18:15] <Keybuk> and why use something that makes it hard when there are easy alternatives that are just as powerful
[18:16] <Keybuk> I also like that bzr is written in a simple language like Python
[18:16] <Keybuk> which makes it delightfully extendable
[18:16] <Keybuk> there's bzr plugins to give you git-like commit workflow
[18:16] <Keybuk> and bzr plugins to add rebase, etc.
[18:17] <Keybuk> but because they're written as extensions to the bzr code, they feel like you're still using bzr
[18:17] <Keybuk> they still work the same way