.This is a pure Perl module I wrote which provides a Perl interface to the CVS client.
This document is a simple introduction to running both a CVS client and a CVS server under Windows.
This is for when you wish to administer CVS yourself, and for using VCS::CVS.
I discuss installing the most recent stable version of CVS for NT because I refuse to install beta software.
I assume you want to start with a minimal system, which means not worrying about web, email and newsgroup servers, nor GUI-based cvs clients.
At http://www.cvsnt.org/ there is a file cvsnt_1.11.1.3.exe (3,342,042 bytes) which is a cvs server. I'm running this.
I log on to WinNT as Administrator, and will be running all Perl scripts under the same account. This means the cvs client will always be run by Administrator.
I do not want CVS to accept connections from anyone on the internet, so the TCP/IP option in the server is disabled. See below.
I use e:\Repository. A simple name means it is simple to back up.
This directory will contain a number of sub-directories. One will be called CVSROOT, and there will be one for each module you import into CVS.
Or wherever. I have not worked out why CVS for NT needs this, since $HOME/.passwd is not used.
I could not get this to work after trying quite a few variations, so forget it.
Install it into, say, d:\Program Files\CVS for NT.
I guess it's better than a purple turkey...
The window says CVS NT Service 1.11.1.3 rc1 (Build 53).
A moment later the Stop button will become clickable.
The client should connect to the server, and output
Concurrent Versions System (CVSNT) 1.11.1.3 rc1 (Build 53) (client/server)
followed by 7 lines of other stuff.
We're up and running!
Firstly decide where you are going to work.
I use e:\Sandbox.
We need to test CVS by doing this:
So here we go.
We'll do all this manually at first, just to ensure it works.
Later we'll replicate it with a Perl program.
But wait!
By default, CVS corrupts binary files. This really pisses me off.
So, restrict yourself to text files until you've studied this problem.
shell>e:
shell>cd \
shell>cd sandbox
shell>md apache
shell>cd apache
shell>md cgi-bin
shell>md htdocs
shell>copy d:\apache\cgi-bin\env.cgi apache\cgi-bin
shell>copy d:\apache\htdocs\index.html apache\htdocs
shell>e:
shell>cd \sandbox\apache
shell>cvs import -m "Initial version" apache start-tag vendor-tag
Your output should look like:
cvs import: Importing e:/Repository/apache/cgi-bin
N apache/cgi-bin/env.cgi
cvs import: Importing e:/Repository/apache/htdocs
N apache/htdocs/index.html
No conflicts created by this import
Notice how we had to be in sandbox\apache to do this.
We do this by checking out the apache module into a slightly different directory and running diff.
shell>cd \sandbox
shell>md test
shell>cd test
shell>cvs checkout apache
Your output should look like:
cvs checkout: Updating apache
cvs checkout: Updating apache/cgi-bin
U apache/cgi-bin/env.cgi
cvs checkout: Updating apache/htdocs
U apache/htdocs/index.html
shell>diff -rs \sandbox\apache \sandbox\test\apache
Notice we are not using the 'cvs diff' command yet, but
rather are running the GNU for DOS diff. See:
http://savage.net.au/Ron/GNU-Utils.html for downloading help.
The -r option means recursive. The -s option means report identical files.
Your output should look like:
Only in \sandbox\test\apache: CVS
Only in \sandbox\test\apache/cgi-bin: CVS
Files \sandbox\apache/cgi-bin/env.cgi and
\sandbox\test\apache/cgi-bin/env.cgi are identical
Only in \sandbox\test\apache/htdocs: CVS
Files \sandbox\apache/htdocs/index.html and
\sandbox\test\apache/htdocs/index.html are identical
By this time, you're laughing (hysterically).
It's easy to import a module and then start editing it. But the copy you've just imported is not the copy you want to be editing.
You must run cvs to checkout a new copy, and edit that copy.
So that's what we turn to next.
Time to edit a file, check it in, and run 'cvs diff'.
shell>e:
shell>cd \sandbox
shell>rd /q /s apache
Notice how we can wipe out the previously imported
apache module now that we have faith in CVS.
shell>rd /q /s test
shell>cvs checkout apache
Do something distinctive. If we assume lines 1 .. 4 contain comments, then add these lines after line 5:
#
# Author:
#<tab>Ron Savage.
shell>cd \sandbox\apache\cgi-bin
shell>cvs commit -m "Added author" env.cgi
Your output should look like:
Checking in env.cgi;
e:/Repository/apache/cgi-bin/env.cgi,v <-- env.cgi
new revision: 1.2; previous revision: 1.1
done
Notice how we were in apache\cgi-bin to do this. Actually, this would work too:
shell>cd \sandbox\apache
shell>cvs commit -m "Added author" cgi-bin\env.cgi
shell>cvs diff -r1.1 -r1.2 env.cgi
Your output should look like:
Index: env.cgi
===================================================================
RCS file: e:/Repository/apache/cgi-bin/env.cgi,v
retrieving revision 1.1
retrieving revision 1.2
diff -r1.1 -r1.2
4a5,7
> #
> # Author:
> # Ron Savage.
It works!
This document is at http://savage.net.au/Perl/html/configure-cvsnt.html
You'll find a self-expanding version of CVS for NT at http://www.cvsnt.org/
Corresponding docs are at http://www.devguy.com/fp/cfgmgmt/cvs/cvs_admin_nt.htm
I do not use WinCVS. There are various versions of WinCVS at http://www.cvsgui.org/download.html
There are various items at http://www.cvshome.org
Good information on CVS-related mailing lists, etc is at http://www.cvshome.org/communication.html
Another Perl module, VCS (unrelated to VCS::CVS), which interfaces to cvs is at http://www.astray.com
Ron Savage .
Home page: http://savage.net.au/index.html
This version disguises my email address.
Original version.
Australian Copyright © 2002 Ron Savage. All rights reserved.
All Programs of mine are 'OSI Certified Open Source Software';
you can redistribute them and/or modify them under the terms of
The Artistic License, a copy of which is available at:
http://www.opensource.org/licenses/index.html