[Condor-devel] condor git repositories would benefit from recompression/gc

[Jim Meyering <jim@xxxxxxxxxxxx>]

Hello,

I cloned the two repositories per instructions here:

    http://ben.versionzero.org/wiki/Using_Git_to_get_the_Code

It took a surprisingly long time to download.
At 238MB, condor's .git/ directory seemed to be too large for the
size of the code.  Typically, git's internal compression does much
better than that.  Let's see what happens when I recompress locally...

    $ du -sh .git; git gc --aggressive; du -sh .git
    238M    .git
    105M    .git

Wow.  Using git's garbage-collection removed more than 130MB of bloat.
(fyi, I've seen such bloat -- much worse, actually -- as a result
of an initial conversion from cvs)

That means everyone who clones the source is currently downloading
more than twice as much data as necessary.  If someone runs the above
commands on the server, first-time cloners will thank you ;-)

Let's do the same for the condor_docs/ repository.
It's smaller, but compressing it reduces it to 1/7th(!) of the original size:

    $ du -sh .git; git gc --aggressive; du -sh .git
    48M     .git
    6.4M    .git

While local git commands are very snappy, you will find that they're
even snappier when you can reduce its working set size by so much.