Thoughts / Misc: March 2007 Archives

I'm running a script to gather SVN details for a graph I'd like to make. I've done this before in perforce and it takes about 5-10 minutes to run. HOURS! HOURS this has run in subversion. Compare:

Subversion:

% time svn diff -r1743:1744 https://svn.example.com/project/trunk > /dev/null
real 0m8.183s
user 0m0.179s
sys  0m0.049s

Perforce:

% time p4 describe 3037 > /dev/null
real 0m0.867s
user 0m0.039s
sys  0m0.091s

Am I doing something horribly wrong? All the doco I can find says that this is the way to get the diffs for a single revision, yet... 10x slower? Hell, CVS is faster!

ADDED:

Some have suggested that it is SSL negotiation. For that I offer the following:

% svn up
At revision 2761.
% time svn -u st -N > /dev/null    
real    0m2.681s
user    0m0.119s
sys 0m0.045s
% time svn diff -r2759:2760 https://svn.example.com/project/trunk > /dev/null
real    0m9.189s
user    0m0.187s
sys 0m0.055s

A 3x increase in time for a 20 line diff??? Nuh uh. Not SSL.

Currently:

• There are 54 authors using hoe.
• Contributing 102 projects to the ruby community!!!
• Hoe is used in 7.18% of all gems!

Hot damn!

2) I should note that with 50+% Texan blood, I'm legally allowed (some might say obliged) to use the phrase "Yee Haw" and its variants.

Matthew King for activemdb!!! YAY Matthew!!!

Other contenders were:

• John Nunemaker with twitter.
• Aaron Patterson with his 4th (!!) hoe'd gem, daapclient (which only missed it by about 30 minutes).

Hoedown reports that we're at 99 hoe'd gems from 52 authors. We've been there since Thursday when Seattle.rb's very own Mike Mondragon published mms2r.