| crab | you can do it pretty easily in p4 |
| fynn | Hey. I vaguely recall there's a way to ignore working-copy subdirectories in a local way. |
| fynn | I.e. it would work on my working copy, but won't be propagated to any clones. |
| fynn | (so it's not by adding a .gitignore) |
| crab | .git/info/exclude |
| crab | takes the same kind of pattern(s) as .gitignore |
| fynn | crab: thanks, that was it. where is it in the mans/docs, btw? |
| tekky | sitaram: heya :) |
| sitaram | hi! |
| sitaram | tekky: ^ |
| tekky | you have a good weekend? (showing my ignorance -- but dont know if you celebrate this time of year :P) |
| sitaram | tekky: absolutely great, but nothing to do with religion (25th anniv reunion of my college class -- yeah I'm ld ;-) |
| sitaram | old* |
| tekky | lol congrats |
| sitaram | and nothing to do with the new year... not yet anyway :) |
| tekky | I just got my inet hooked up :P |
| fynn | was there any change in 1.6.6 wrt how work trees are handled? |
| tekky | 16M down, 4-8M up :) |
| Action: sitaram is jealous... he can get 10% of that (assuming bps not Bps) on a really good day |
| tekky | sitaram: Download Speed: 16433 kbps (2054.1 KB/sec transfer rate) ---- Upload Speed: 5484 kbps (685.5 KB/sec transfer rate) |
| tekky | less confusion |
| fynn | I'm using the following command: GIT_WORK_TREE=$HOME/rawdata GIT_DIR=$HOME/rawdata/.git git reset --hard |
| fynn | with earlier versions, it worked fine |
| fynn | with 1.6.6, I get: "fatal: hard reset requires a work tree" |
| fynn | which only succeeds if I run the command from within $HOME/rawdata |
| sitaram | fynn: try not using both GIT_WORK_TREE *and* GIT_DIR maybe? (I've never used both; the current directory was always one of them) |
| fynn | sitaram: $ GIT_WORK_TREE=$HOME/rawdata git reset --hard |
| fynn | fatal: Not a git repository (or any of the parent directories): .git |
| fynn | again, the same command works when I'm in ~/rawdata |
| fynn | it also occurs when I use the --work-tree argument :/ |
| fynn | this could be a bug. |
| sitaram | fynn: gitolite easy install *depends* on "cd to bare repo; GIT_WORK_TREE=somewhere-else git subcommand" and I just tested with 1.6.6 |
| sitaram | fynn: and it works fine |
| sitaram | (haven't tried reset --hard; my commands are add, diff, commit) |
| fynn | well, it doesn't work here. |
| fynn | OK, this is most definitely a version-related problem: |
| fynn | $ GIT_WORK_TREE=$HOME/rawdata/ GIT_DIR=$HOME/rawdata/.git /usr/local/git-1.6.6/bin/git reset --hard |
| fynn | fatal: hard reset requires a work tree |
| fynn | $ GIT_WORK_TREE=$HOME/rawdata/ GIT_DIR=$HOME/rawdata/.git /usr/local/git-1.6.4/bin/git reset --hard |
| fynn | HEAD is now at 77ec73f... |
| sitaram | fynn: my bare repos are not called ".../.git" and they *do* have a "core.bare = true". I'm not saying that 1.6.6 didn't change -- I'm trying to narrow it down so you know how to report it :) |
| fynn | sitaram: yeah, thanks. I'm probably going to report it. |
| sitaram | fynn: aha... reset --hard doesn't work! |
| sitaram | add, diff, commit, do! |
| fynn | yay, I found a genuine bug |
| fynn | sitaram: should I just send an email to the mailing list? |
| sitaram | wait |
| sitaram | testing with checkout -f also (which is the same thing in this special case) |
| sitaram | yup; that works fine, but reset --hard doesn't. |
| sitaram | in a way this makes sense... reset is about changing HEAD itself |
| fynn | sitaram: cool, sending that email. |
|
|
| sitaram | fynn: wait |
| sitaram | fynn: well you can send it but I have a link that might shed some light on this |
| sitaram | fynn: http://permalink.gmane.org/gmane.comp.version-control.git/104457 |
| sitaram | fynn: there's a fundamental difference in what a "reset" does versus what a "checkout" does; and I think -- even though the behaviour changed -- there would be an explanation for why. So ask away, and I'd be interested in the response too |
| fynn | sitaram: *nod*, sending. |
| fynn | sent. |
| fynn | sitaram: thanks for making the effort to confirm this. |
| sitaram | you're welcome! |
| sitaram | (I mean it may not be a *bug* per se, maybe an intentional change...) |
| chuck | why doesn't git automatically setup tracking branches for all remote branches? |
| adamm | because many times I don't want it to do that |
| fynn | chuck: what if the remote has 1,000 branches, and 999 of them only interest the remote user? |
| fynn | under most distributed workflows, you're usually not automatically interested in a branch created on the remote; the opposite, in fact. |
| developish | i have a file that is tracked in my repo, but i want to ignore future changes to it, how do i do that? |
| sitaram | developish: the first para of "man git-ignore" will tell you |
| ArjenL | man gitignore (or 'git help ignore' :-) |
| Gitbot | ArjenL: the gitignore manpage can be found at http://git.or.cz/man/gitignore |
| ravionrails | i have made local server on my LAN and push my projects . i get only .git and nothing else there . why? |
| tomoj | ok, I am trying to understand the "you asked me to pull without telling me which branch you want to merge with, and 'branch.nullsfix.merge'..." error |
| tomoj | is this something new because I upgraded? or did I somehow wind up in some situation I just hadn't ever been in before? |
| Ilari | ravionrails: $faq non-bare |
| Gitbot | ravionrails: Pushing to non-bare repositories is discouraged. See http://git.or.cz/gitwiki/GitFaq#non-bare |
| developish | now how do i access the "git update-index --assume-unchanged documentation." |
| Ilari | developish: 'man git-update-index'? |
| Gitbot | developish: the git-update-index manpage can be found at http://git.or.cz/man/git-update-index |
| tomoj | ooh, I see the problem |
| tomoj | didn't realize I was on the 'nullsfix' branch |
| developish | great i didn't know how to use the manpages for all those commands |
| fynn | sitaram: still awake? |
| alrock | hi, I'm using git with cygwin and whenever I "git push" to the server, it changes the permissions of my files in the repository |
| sitaram | fynn: I'm at work, it's 2:46pm! |
| fynn | no it's not, it's 1:16am |
| alrock | it removes the execute/modify etc permissions which creates some problems with the app |
| sitaram | (and there's not a meeting in sight, so I can't really sleep! ;-) |
| fynn | sitaram: http://marc.info/?l=git&m=126207591212169&w=2 |
| fynn | I'm going to ask why that patch is necessary. |
| sitaram | bot, tell me 952dfc6 |
| Gitbot | [git 952dfc6]: http://tinyurl.com/yf3okte -- reset: improve worktree safety valves |
| fynn | sitaram: commit is here: http://github.com/git/git/commit/952dfc6944b29582482ff50a85c04879406c06ba |
| sitaram | fynn: the correct question is "why does this not honor GIT_WORK_TREE", I think |
| Rhonda | When I created a new branch locally that I want to push to a remote, a "git push origin newbranch" is sufficient. But that doesn't add the tracking of the branch to .git/config. What am I missing, is there a simple option that I haven't found? |
| sitaram | and you might wanna mention that "git checkout -f" does work... |
| raboof | i'd like to work on the 'mob' branch of 'http://repo.or.cz/w/jack_mixer.git'. |
| raboof | so i should just clone ssh://repo.or.cz/srv/git/jack_mixer.git and then switch to the 'mob' branch, right? |
| fynn | sitaram: thanks, I'll go give them a piece of my mind! |
| bob2 | raboof: clone, then make your own branch and work on that |
| alrock | anyone? :/ |
| selckin | alrock: git is not a deployment tool |
| raboof | bob2: there's work in the 'mob' branch that's not in master yet and that i'd like to build on |
| bob2 | raboof: branch from mob then :) |
| sitaram | fynn: erm no I didnt mean that (piece of my mind!) |
| sitaram | fynn: as I said, there is a difference in intent between reset and checkout, and am wondering if that is the reason for this in some way that I do not yet understand completely |
| raboof | bob2: how? i tried 'git checkout mob', but that gets me 'error: pathspec 'mob' did not match any file(s) known to git'. 'git branch mob' works, but it looks like that makes a new local branch instead of switching to the existing mob branch. |
| fynn | sitaram: too late, I already posted the most inflammatory flame on the git list! |
| raboof | ('git branch mob' and then 'git checkout mob', that is) |
| sitaram | fynn: :) |
| fynn | I insulted Junio, Linus, and sweet baby Jesus |
| selckin | raboof: git branch -a |
| bob2 | raboof: git checkout -b mob origin/mob |
| sitaram | fynn: and two of tehm had birthdays recently, AFAIK :) |
| sitaram | them* |
| fynn | two... or are they really one?! |
| bob2 | though I think checkout has a magicer way to do that |
| fynn | sitaram: (more seriously, I think if you specify both GIT_WORK_TREE and GIT_DIR, the reset should work) |
| raboof | bob2: ha, looks like 'git checkout remotes/origin/mob' was what i was looking for - thanks |
| raboof | selckin: thanks ;) |
| bob2 | raboof: that doesn't make a branch you can commit to |
| bob2 | raboof: or does 'git branch -a' show that it did? |
| alrock | selckin: so what do you suggest I should do? |
| fynn | sitaram: (that patch was meant to prevent .git from getting accidentally mangled; if you specify both the above, and the repo is non-bare to boot, there's no way that I can see that this patch improves repo protection) |
| raboof | bob2: oh, right, i'm now at '(no branch)' |
| selckin | alrock: autotools seem popular for installing stuff |
| bob2 | raboof: I think you want the command I suggested - it creates a new (local) branch based on 'mob' and checks it out |
| raboof | bob2: yeah, that looks better |
| sitaram | fynn: agreed... |
| sitaram | fynn: it should just honor GIT_WORK_TREE really |
| alrock | selckin: I'm using it like http://toroid.org/ams/git-website-howto so that I can commit locally then push it to server to update the live site/code, it's just one server |
| selckin | alrock: bad advice is bad |
| alrock | how can I setup an alias for "ssh://name@server/" part? |
| alrock | I don't want to store the repo path in the alias |
| alrock | just the username@server |
| Ilari | alrock: insteadof lines (I don't remember the syntax outright)? |
| alrock | Ilari: what? |
| Ilari | alrock: Search man git-config for "insteadOf". |
| Gitbot | alrock: the git-config manpage can be found at http://git.or.cz/man/git-config |
| alrock | ah, that's it. thanks |
| GDG | how can I roll back to a previous file version with git? in svn I use svn up -r<rev>. |
| selckin | checkout |
| GDG | selckin, but checkout to which version? |
| selckin | the one you want |
| GDG | how can I get the version |
| GDG | sorry |
| selckin | git log? |
| bob2 | GDG: you can tell it the version in lots of different ways |
| GDG | I tried git log |
| GDG | but I am never seeing a version id |
| GDG | (sorry I am a svn user) |
| nevyn | 0f8d51b |
| nevyn | is a recent version of a tree I'm tracking |
| nevyn | you should see a bunch of stuff like that. |
| GDG | commit 581e06110a00669da1b54b82225fcac357d8f154 |
| GDG | I see this |
| GDG | is this my version id? |
| selckin | yes |
| GDG | it is uge |
| GDG | or better it is an hash |
| selckin | it's a sha1 |
| GDG | ok |
| selckin | there are other ways to specify versions |
| GDG | which are |
| bob2 | you can use just a prefix |
| bob2 | man git-rev-parse |
| GDG | I just need 1 step back |
| Gitbot | bob2: the git-rev-parse manpage can be found at http://git.or.cz/man/git-rev-parse |
| wereHamster | you can just use the first four or five characters of the hash |
| bob2 | HEAD^ |
| GDG | what is HEAD^ ? |
| bob2 | the previous revision |
| ambuli | what the equal command for svn update ?? in git |
| nevyn | ambuli: git pull |
| nevyn | ambuli: but git pull is syntactic sugar around a git fetch and git merge |
| nevyn | ambuli: so you can do a 2 phase update if you want which will firstly fetch the upstream changes to the indexx |
| nevyn | then let you examine the upstream changes and integrate them into your tree |
| ludde2 | hi nevyn |
| ambuli | nevyn, what is the command i have use for both ?? |
| ludde2 | nevyn: do i know you? :p |
| nevyn | ambuli: git pull |
| ludde2 | i guess not |
| ambuli | nevyn, so git pull does both phase is it ?? |
| ludde2 | you seem to be a different nevyn |
| nevyn | possibly |
| Ilari | The changes are not fetched into the index. They are downloaded to refs. |
| nevyn | ambuli: yes it does.. |
| ludde2 | you're not from sweden, are you |
| nevyn | Ilari: my bad. you're correct.. |
| nevyn | ludde2: no I'm not |
| nevyn | I'm from australia |
| ambuli | nevyn, so git fetch is for phase 1 and git merge is for phase 2 if i understand correctly ?? |
| nevyn | ambuli: yes |
| ambuli | nevyn, Ah.. cool. Thanks nevyn |
| nevyn | ambuli: and git pull does both. |
| nevyn | but I only ever use git pull where I'm not doing work |
| nevyn | so if I'm just using git as a tracking tool for an upstream library or something I'll git pull |
| nevyn | ambuli: or if you work in a branch and keep master just to track upstream then that works too. |
| ambuli | neunon, ok |
| Juerd | I have two completely different repositories, one local, one remote. I want to add the local files to the remote repo. Is it possible to retain the changes history? |
| nevyn | they don't have a common root at all? |
| Juerd | No |
| ambuli | nevyn, what about --amend option ?? |
Popular searches: