[cvsnt] Recursive chacl does not recurse properly and hangs on deep directories.

[Mike Wake]

G'day Tony,

ACL testing with v2.0.58d+(cvs lsacl fix and chown fix) continues :)

If I do this
cvs chacl -R -u my_team -a read,write,create,tag

in a sandbox containing a number of levels of directories all of which I 
own.

Only the top level directory gets all of the acls set for the user "my_team"

The subsequent directories only get the "read" acl set.

Here is the result of cvs lsacl on one of these directories
Directory: .
Owner: mwake

<default>
	read

user=my_team
	read



If I do this on deep directories ( I haven't isolated how deep yet at 
least 7 dirs deep )

On the first run only some of the top level directories have their acls 
changed at all. Only the top directory gets changed correctly.

On the second run the client never finishes communicating.  When I got 
bored with waiting and interupted it I get

cvs chacl: warning: unrecognized response `setting ACL for directory .' 
from cvs server
cvs chacl: warning: unrecognized response `setting ACL for directory 
CVS_info' from cvs server
cvs chacl: warning: unrecognized response `setting ACL for directory 
CVS_info/CVS-BestPractices' from cvs server
cvs chacl: warning: unrecognized response `setting ACL for directory 
CVS_info/StreamedLines' from cvs server
cvs chacl: warning: unrecognized response `setting ACL for directory 
CVS_info/binaries' from cvs server
cvs chacl: warning: unrecognized response `setting ACL for directory 
CVS_info/images' from cvs server
.
.
.
.
cvs chacl: warning: unrecognized response `setting ACL for directory 
SCM_CM/lbd/dev/src/rtsubs/rs00/i686' from cvs server
cvs chacl: warning: unrecognized response `setting ACL for directory
SCM_CM/lbd/dev/src/rok' from cvs server
cvs [chacl aborted]: received interrupt signal

Cheers
Mikew