cvs checkout -d & Max-dotdot

[John Daniel Doucette]

Hello all,

I have been doing development on a project for a couple of years and have 
been using CVS to control source.  Up to now all repository access has been 
on my local machine, i.e. :local:...  I am required to move the repository 
to a Linux server, and make it available to the company's LAN using the 
:pserver: client/server protocol.  I have moved the repository 
successfully, and can access it from different XP/Win200 
machines.  However, the repository contains hundreds of versions of 
makefiles that use the -d <path> option with the checkout command to check 
out sources in other than the current directory. I.e. there are many lines 
with commands like "cvs checkout -d ../../../stuff".  This apparently 
doesn't work between client/server.  An actual error message I get is:


mkdir: created directory `../../../../../top/middle/bottom/working'
cvs server: protocol error: `../../../../../top/middle/bottom/working' 
contains more leading ..
cvs [server aborted]: than the 0 which Max-dotdot specified

Is this a server bug or client bug? It seems like a client bug, since 
checkout -d should only affect the destination of the checked out files on 
the client machine, and has nothing whatsoever to do with the repository 
path of the modules being checked out.   I can easily change references in 
the latest versions of all makefiles to use a cd ../../../../.. followed by 
checkout, but this doesn't really help me if someone wants to build a old 
version.  I can't feasibly change all versions of the makefiles!  I have 
found reference in the CVS client/server protocol documentation to 
"Requests" that can be made of the server.  One of the requests is the 
Max-dotdot option.  What would happen if I used Telnet to send this request 
to the server with a reasonable value after I have logged in and before I 
have checked out and run my makefile?  Would the Max-dotdot value "stick 
around" after the telnet communication and be applied when I run the 
makefile?  Does anyone know what this theorized telnet session would look 
actually like?  The cvs client/server protocol documentation is not super 
clear to me.  On the Linux Red Hat box I am running: Concurrent Versions 
System (CVS) 1.11.1p1 (client/server). On my PC I am running: Concurrent 
Versions System (CVSNT) 1.11.1.3 (Build 72) (client/server).  Are there any 
environment variables or command-line options that can be added to the 
xinetd.d/cvspserver service file to affect it's Max-dotdot setting? Any 
ideas/thoughts would be appreciated. Thanks.

===================================
John Daniel Doucette, Sr. Software Designer

J. J. MacKay Canada Limited
Halifax R & D Office
1046 Barrington Street, 1st Floor
Halifax, Nova Scotia
Canada
B3H 2R1

Voice: 902.423.7727  x222
Fax:     902.422.8108
Web:    www . jjmackay . com
Email:  john . doucette @ jjmackay . com
===================================