AMC's git FAQ

["Alan M. Carroll" <am...@network-geographics.com>]

* Note - this contains only questions because I have no idea how to do any of these things despite hours of online research on git. But I certainly ask them of myself frequently. *

1) How do I get the revision control status of a specific file? I want to know if the file is in the repository or not, and if in, whther it has been modified in my current checkout and the last time / commit it was modified in the repository.

2) If I start making changes on master because it's just a quick little fix, then find out it's not so trivial and I need a branch, how do I put those changes on a branch without losing them?

3) If I end up with a local branch with the same name as a remote branch, how do I make my local branch a tracking branch for the remote one? Especially if I have committed changes on the local branch?

4) How are old branches to be culled? For instance, the ipv6, wccp, and ts-291 branches no longer make sense. In SVN, we could move those to an archive directory so they could still be located if needed, but wouldn't create clutter. What happens to commits on a branch that is deleted in git?

Re: AMC's git FAQ

[Leif Hedstrom <zw...@apache.org>]

On 1/28/12 5:43 PM, Alan M. Carroll wrote:
> * Note - this contains only questions because I have no idea how to do any of these things despite hours of online research on git. But I certainly ask them of myself frequently. *
>
> 1) How do I get the revision control status of a specific file? I want to know if the file is in the repository or not, and if in, whther it has been modified in my current checkout and the last time / commit it was modified in the repository.

git log for example. git status also shows stuff, but only for files in the 
repo. git log on a file that's not in the repo should show nothing.

>
> 2) If I start making changes on master because it's just a quick little fix, then find out it's not so trivial and I need a branch, how do I put those changes on a branch without losing them?

   git checkout -b bug/TS-666


That should both create the bug/TS-666 branch, and bring all your changes 
with you over as you "check it out".
>
> 3) If I end up with a local branch with the same name as a remote branch, how do I make my local branch a tracking branch for the remote one? Especially if I have committed changes on the local branch?

It's quite likely your local branch (e.g. 3.0.x) would have the same names 
as the remote (origin/3.0.x). If they were not branched off from the remote, 
you can supposedly (with git v1.7 and later, I have not tried it), do:

|   git branch --set-upstream foo origin/foo|



But in general, to create the 3.0.x branch, you want to do e.g.

   git checkout origin/3.0.x
   bit checkout -b 3.0.x


or probably better

|   git checkout --track -b 3.0.x origin/3.0.x|

>
> 4) How are old branches to be culled? For instance, the ipv6, wccp, and ts-291 branches no longer make sense. In SVN, we could move those to an archive directory so they could still be located if needed, but wouldn't create clutter. What happens to commits on a branch that is deleted in git?
>

I think once you "remove" the branch on the remote, it's gone forever (but I 
might be wrong). I sent a different email, but I'd like to start off by 
cleaning up the branches that we have (remove the ones that are bogus, and 
perhaps rename the rest). We should come up with some better naming scheme 
for branches too. I personally like e.g. "bug"  as a subtree for all bugs. 
This is probably a discussion to make on a different thread though.

Note that branches in git are much (much) cheaper than in SVN. They won't 
take up endless amounts of storage on the "local" repo for starters. 
Switching between branches is virtually instant. I think to avoid clutter, 
the best bet is to try to come up with a hierarchy, and honestly, prune 
branches that are no longer necessary.


I hope that helps?

-- Leif