Hi,
My friend is using CVS 1.11.17 (as a server) on SunOS (Kernel version: Generic_108528-24 & Processor: sparc) machine and WinCVS 2.0 GUI. The CVS passwd file contained only one row and had system user name present in it (third field). All developers were using the same name (jim) for check in & check out. jim:xp88dsdkkllyydk:cvstest This was bad because this way, we will not be able to know who did what. Anything that anyone does will be logged in CVS history as user jim’s changes. Please correct me if I’m wrong. So I thought of adding all the usernames to the CVSROOT passwd file. When I added one more entry to the CVS passwd file (with no system user name present in the 3rd field) and tried logging in using WinCVS, I was able to successfully login and checkout the code but I couldn’t check in the modified code. It gave the following error: cvs commit: Empty password used – try ‘cvs login’ with a real password cvs [commit aborted]: authorization failed: server XYZ rejected access to /usr/local/data/ for user cvstest ***** CVS exited normally with code 1 ***** Contents of CVS passwd file: --------------------------- jim:xp88dsdkkllyydk:cvstest fish:rKa5jzULzmhOo --------------------------- What I understood from the official CVS guide is that CVS passwd file CAN contain any imaginary username (one that is not present in Unix /etc/passwd file). Even if user 'fish' is not a system user name, it should be able to login and perform all CVS operations. We are able to perform all CVS operations only in the following cases: 1. First field of CVS passwd file contains a system user name with NO third field. 2. First field of CVS passwd file contains an imaginary name and third field contains a system user name. Can anyone please explain me this behaviour? Regards, Gaurav -- View this message in context: http://www.nabble.com/Non-system-user-not-able-to-check-in-tp23020300p23020300.html Sent from the Gnu - Cvs - Info mailing list archive at Nabble.com.
