On Sat, 5 Feb 2005, Greg Sabino Mullane wrote:
> I did it last time. It's been a while now, but I think what I
> did was basically look at all the commit messages from the
> previous release to the current one, and then used a perl script
> to extract everything that looked like a name or an email
> address. Then I manually went through and cleaned things up by
> verifying names, removing duplicates, etc. This relies on the
> actual commiter giving credit to the patcher, but everyone
> here is really good about doing that. :) Not sure if I still have
> the script around, but I can dig it up if it's needed.
I'm trying the following on the archives:
grep From: `find 2004-* 2003-1[12] -type f -name "msg*" -exec grep --silent "^diff " {} \; -print` | \ awk -F:
'{print$3}' | \ sed 's/<\/em>//g' | \ sed 's/</ /' | \ awk '{printf"%s %s\n", $1, $2}' | \
sort-u
The problem with commit logs is that a good portion are just 'reports
from' vs patches ... neither method will necessarily be particularly
accurate :)
----
Marc G. Fournier Hub.Org Networking Services (http://www.hub.org)
Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664