Tracking few files among many

Tracking few files among many

[Peter Eriksen]

Hello,

There's something I can't figure out.  I'm tracking a few
configuration files in $HOME, but some operations are really
slow.  Let's take git-status as example: 

~ > git-status >laaangsom.txt
~ > wc -l laaangsom.txt
25875 laaangsom.txt
~ > cat laaangsom.txt
#
# Changed but not updated:
#   (use git-update-index to mark for commit)
#
#       modified: .gaim/blist.xml
#       modified: .gaim/prefs.xml
#       modified: .mozilla/firefox/s4q22693.default/prefs.js
#
#
# Untracked files:
#   (use "git add" to add to commit)
#
#   [ A lot of untracked files. See the line count above. ]

What is going on?  This really doesn't seem like the wanted
behavior.  Have I missed something?  I'm using
46774a81f9d6ca4d230d33757afe9dd07bfe398b  (GIT 0.99.9).

Regards,

Peter

Re: Tracking few files among many

[Andreas Ericsson]

Peter Eriksen wrote:
> Hello,
> 
> There's something I can't figure out.  I'm tracking a few
> configuration files in $HOME, but some operations are really
> slow.  Let's take git-status as example: 
> 
> ~ > git-status >laaangsom.txt
> ~ > wc -l laaangsom.txt
> 25875 laaangsom.txt
> ~ > cat laaangsom.txt
> #
> # Changed but not updated:
> #   (use git-update-index to mark for commit)
> #
> #       modified: .gaim/blist.xml
> #       modified: .gaim/prefs.xml
> #       modified: .mozilla/firefox/s4q22693.default/prefs.js
> #
> #
> # Untracked files:
> #   (use "git add" to add to commit)
> #
> #   [ A lot of untracked files. See the line count above. ]
> 
> What is going on?  This really doesn't seem like the wanted
> behavior.  Have I missed something?

Apart from the fact that git tracks objects using sha1-hashes, no.

However, hashing +25000 files takes quite some time even on a fairly 
quick computer. I also imagine that some of those files are quite large, 
so that doesn't really help.

If you really (really, really) want to use git to track configuration 
file changes in your homedir, I'd suggest creating a separate directory 
to keep the real files in and then symlink to those from their usual 
locations.

OTOH, since you *know* git-status (precisely because it looks for files 
not added to the index) to be slow, you should use git-diff* instead. I 
imagine you know what files you're tracking anyways since it's just a 
subset of 25000-something.

-- 
Andreas Ericsson                   andreas.ericsson@op5.se
OP5 AB                             www.op5.se
Tel: +46 8-230225                  Fax: +46 8-230231

Re: Tracking few files among many

[Peter Eriksen]

On Sun, Oct 30, 2005 at 04:29:27PM +0100, Andreas Ericsson wrote:
> Peter Eriksen wrote:
> >Hello,
> >
> >There's something I can't figure out.  I'm tracking a few
> >configuration files in $HOME, but some operations are really
> >slow.  Let's take git-status as example: 
...
> OTOH, since you *know* git-status (precisely because it looks for files 
> not added to the index) to be slow, you should use git-diff* instead. I 
> imagine you know what files you're tracking anyways since it's just a 
> subset of 25000-something.

I understand now, I just didn't think about this behavior before.
What I could do is add all my files to .gitignore, but that would
be insane.

Perhaps I was just confused about it calling those files "not tracked".
I know, I know, it was changed recently to this exactly because it
was confusing before.

Thanks for clearing that up for me.

Peter