Re: Ignoring files known to CVS, when updating

[Todd Denniston]

Marc-André Boucher wrote:
> Is there a way to ignore files that are already known to CVS? The problem is I
> want to deploy my project to multiple servers/environments/sites, but I don't
> want to override the configuration files every time I create my deployment
> package.
> I have a single directory with my project which I update using CVS, and
> multiple directories with the configuration files of each of the
> servers/environments/sites where I deploy.
>
> I would like to run a single "cvs update" command to keep the project
> up-to-date, but ignoring the config files so that I don't have to modify them
> everytime they are committed in CVS.
>
> I have modified them directly on the deployment server, and they merge
> everytime I update the project, which is ok, but if the configuration files
> are committed for some changes, I don't want them to merge : I'm gonna modify
> them manually if I have to.
>
> I know it seems like a software deployment problem, but the CVS part
> is the heart of it. I have found other ways to do this, but they're too much
> of a "patch" than a real solution!
>
> Thanks again for your time. If there's a way to do this, that would spare me
> some problems!


I think you can do this with the modules file.

What follows is a paste-able script to replicate what I did to prove it to 
myself.



cd /tmp/
mkdir testme
cd testme/
mkdir repo
cvs -d /tmp/testme/repo/ init
mkdir repo/myroot
cvs -d /tmp/testme/repo/ checkout myroot
cd myroot/
mkdir dir1
mkdir dir2
mkdir dir3
for i in `seq 1 5`;do date > dir1/common$i;done
for i in `seq 1 5`;do date > dir2/com$i;done
for i in `seq 1 5`;do date > dir3/specific$i;done
cvs add dir*
cvs add dir*/*
cvs commit -m "dont ask"
cd ..
cvs -d /tmp/testme/repo/ checkout CVSROOT
#make the following mod.
--- CVSROOT/modules     8 May 2007 10:08:43 -0000       1.1
+++ CVSROOT/modules     8 May 2007 20:09:18 -0000       1.2
@@ -24,3 +24,6 @@
 # character to interpose another module into the current module.  This
 # can be useful for creating a module that consists of many directories
 # spread out over the entire source repository.
+
+specific2  -a myroot/dir1 myroot/dir2 myroot/dir3/specific2
+

cvs commit -m "dont ask" CVSROOT/modules
mkdir codir
cd codir/
cvs -d /tmp/testme/repo/ checkout specific2
for i in `seq 1 5`;do date >> \
    /tmp/testme/myroot/dir3/specific$i ;done
cvs commit -m "dont ask" /tmp/testme/myroot/dir3/
cd myroot/dir3/
cvs update
ls -ltrh ;echo "should only see specific2"


IIRC the reason this works is the CVS/Entries file in dir3 only contains the 
entry for specific2 and so it does not look for others unless you update with 
the '-d' option.


BTW, using the -a was not obvious to me until I understood that it is telling 
cvs to replace the module name on the command line with exactly what comes 
after the module name in the modules file ... i.e.
specific2  -a myroot/dir1 myroot/dir2 myroot/dir3/specific2
is not the same thing as
specific2   myroot/dir1 myroot/dir2 myroot/dir3/specific2
and I don't understand why.

if you don't want to create  the modules file you could do (at the checkout 
place in the script after I indicate mod the modules file):
cvs -d /tmp/testme/repo/ checkout \
    myroot/dir1 myroot/dir2 \
    myroot/dir3/specific2


--
Todd Denniston
Crane Division, Naval Surface Warfare Center (NSWC Crane)
Harnessing the Power of Technology for the Warfighter