This document attempts to provide in depth details of the means by which jCVS connects to the cvs server.
Another way to debug connections is to use the Checkout Dialog and check all of the tracing checkboxes. This will cause detailed debugging of the server connection and protocols to stderr.
Finally, refer to the common problems page to see if there are any suggestions regarding your specific environment.
Because of this, jCVS provides a means for you to specify that it
should exec() the rsh program to communicate to the cvs server.
To do this, you must un-comment the following line in the
'Properties.txt' file:
The Properties file has an override of this property for the Solaris operating system, so Solaris users should not have to set this property.
In the case of jCVS, we use the 'direct' method to establish
a direct tcp connection to the server with no password
authentication, and no rsh protocol. The server must be started
via inetd or an equivalent. The inetd entry must start the
server with the 'server' argument, as opposed to 'pserver'.
Here is an example entry:
cvss stream tcp nowait time /usr/local/bin/cvs cvs server
NOTE that the process is started with a user id that is NOT 'root'. This is the only level of security that this connection method provides, so select the user carefully.
The default port number is 2402, but this can be controled via the 'Properties.txt' file. Look for property 'jCVS.portNum.direct.default', and company.
To tell jCVS to use the 'direct' method, check the 'Pserver' checkbox, and uncheck the "Use Password" checkbox. This will make the UserName and Password fields disappear, since they are not used with this connection method. Which brings up a very important point.....
THIS CONNECTION METHOD SHOULD ONLY BE USED FOR COMPLETELY ISOLATED LOCAL AREA NETWORKS, WHICH CAN IN NO WAY BE ACCESSED FROM A COMPUTER NOT INTENDED TO ACCESS THE SERVER!!!
You will be hacked if you ignore this advice.