views:

51

answers:

2

I have a bare repository and two working copies - one on my machine, the other on the server.
It turned out that I have to .gitignore a certain file that has to be specific for every machine. Let's call it 'settings.py'. This file is already commited.

I did put 'settings.py' in .gitignore to ignore it. When I now change the file on my machine git status still tells me

modified:  settings.py

I figured out that I have to remove settings.py like this:

git rm --cached settings.py

Then git add ., followed by git commit.

But when I now push this to the bare repo and pull it to the working copy on the server, settings.py is deleted there - which is bad because I have to keep this specific settings.py.

I figured that I just could make a copy of settings.py and put it back in once it is deleted, but I feel like there has to be a better way of doing this.

+2  A: 

Could you rather keep settings.py deleted permanently (both locally and on remote), and:

  • version a setting_remote file for the remote side
  • version a setting_local file for the local side (ie with local specific settings)
  • add a filter driver able to rebuild the right settings.py file on checkout.

alt text

That way, settings.py is kept "private" (non-versioned). But the specific values for each environment are versioned, each set in its own file, without any merging issue.

VonC
A: 

You can tell git to completely ignore changes to tracked files without having to remove them. Use this command:

git update-index --assume-unchanged [FILENAME]

Then if you want to track the file later:

git update-index --no-assume-unchanged [FILENAME]
Outspaced