Git Clone vs cp -R --> WTF?
I knew git was fast, and I even knew it was faster than a lot of plain linux local file operations. Still, this still blew me away:
[code]rsanheim@ares:~/src/personal/oss $ du -hd 0 insoshi/ 26M insoshi/
rsanheim@ares:~/src/personal/oss $ time git clone insoshi/ /tmp/insoshi Initialize /tmp/insoshi/.git Initialized empty Git repository in /private/tmp/insoshi/.git/ Checking out files: 100% (2193/2193), done.
real 0m3.826s user 0m0.251s sys 0m0.658s
rsanheim@ares:~/src/personal/oss $ time cp -R insoshi/ /tmp/insoshi_cp
real 0m9.065s user 0m0.114s sys 0m1.442s [/code] Ok, so a 26 meg repo takes almost three times as long to copy via a recursive cp than a local git clone. Thats a fairly small repo, lets try something bigger:
[code] rsanheim@ares:~/src/relevance $ du -hd 0 rails 75M rails
rsanheim@ares:~/src/relevance $ time git clone rails /tmp/rails2 Initialize /tmp/rails2/.git Initialized empty Git repository in /private/tmp/rails2/.git/
real 0m2.321s user 0m0.151s sys 0m0.465s
rsanheim@ares:~/src/relevance $ time cp -R rails/ /tmp/rails
real 0m7.133s user 0m0.067s sys 0m1.505s [/code]
The rails repo at 75 megs is still ~ 3 times faster.
Obviously, this is not scientific at all, but the point is pretty clear. Git is doing some magic that lets it move files around locally 2 to 3 times faster than a plain copy. From looking at the man page, I would guess it has something to do with git using hardlinks for things in .git/objects when cloning locally. My linux fu falls down a bit here -- what are the ramifications of using hard links versus doing a "real" copy?
(This also makes me want to try out gitbak even more...)