[ILUG] CVS synching

[Brendan Kehoe]

> How do I sync up with this without having to do a full 'cvs checkout' which
> I absolutely refuse to do over a 28.8k modem?  ... All I need is basically the
> little CVS directories in the tree.

You could do something like

cd blah
find * -type d -exec /tmp/newcvs {} \;

The use of `find * ...' with an asterisk is deliberate, so the argument going
into Repository looks correct.  Thus the cd to make sure that you're at the
top of the source tree.

The /tmp/newcvs script is

#!/bin/sh
[ "$1" -eq "CVS" ] && exit 0
cd $1 || exit 1
[ -d CVS ] || mkdir CVS
# 1 - Tell it where the CVS repository as a whole lives, so the cvs client knows
# where to go.  Change :ext: appropriately to your $CVSROOT's approach.
rm -f CVS/Root
echo ':ext:user at site.com:/some/cvsfiles' > CVS/Root
# 2 - Tell it the directory that has this particular subdirectory.
rm -f CVS/Repository
echo "/cvs/cvsfiles/$1" > CVS/Repository
# 3 - For each file, find out the correct version number and make note of it in
# the Entries list.
rm -f CVS/Entries
cvs status * 2>&1 | grep ,v | \
while read line ; do
  # Line format: 
  #    Repository revision: 1.18    /some/cvsfiles/blah/foo/code.c,v
  v=`echo $line|awk '{print $3}'` 
  f=`echo $line|awk '{print $4}'|cut -f1 -d,`
  f=`basename $f`
  echo "/$f/$v/`date`/" >> CVS/Entries
done
exit 0

The cost of doing a `cvs status' on your tree should be relatively cheap even
on a 28.8 connection, in comparison to checking out the full tree.

Note the above assumes the source tree you have is in fact up to date with
what was imported into the CVS repository.

Hope this helps!
B


-- 
Brendan Kehoe

Web page: http://www.zen.org/~brendan/