Johannes Buchner

CVS is based on RCS, which is file based. That means, you don’t really have revisions over multiple files. The “cvs log” command reflects that problem. This outputs a more useful summary: date and commit comment per line.

cvs log > cvslog # puts log in this file
cat cvslog |
grep '^date: ' -A3 | # we assume max 3 lines of comments
while read line; do echo -n "$line |"; done | # all in one line, seperated by |
sed 's/date: /\ndate: /g'| # a line per date
sed 's/[-=]\{2,\}/\n/g' | # removing line seperators
sed 's/^[- |]*//g'| # remove useless - and |
grep -v '^revision '| # there exist some more boring lines
sed 's/;  author:.*; |/;/g'| # we aren't interested in author and id etc.
sed 's/ |$//g'| # remove ending |
sort -u | # sort (by date)
python guniq.py | # show only uniq lines
cat > cvslog.1

You need guniq.py which does the same as the unix command uniq, except that it removes duplicates found in the whole input.

guniq.py:

#!/usr/bin/python
import sys
a = []
while True:
l = sys.stdin.readline()
if l == '':
break
if not a.__contains__(l):
print l,
a.append(l)

The whole thing got me nearly to boot into Windows, because TortoiseCVS/TortoiseSVN is a really cool awesome thingy!