Package management

[Neil Joseph Schelly]

On Tuesday 15 May 2007 10:04, Charlie Farinella wrote:
> How does everyone else do this?

In the Debian world, at least, I generally follow this order of precedence.

1.  If it's outside the distro availability, I look to upgrading.  In this 
case, Python 2.3.4 is pretty outdated, so I'm assuming there are newer CentOS 
distros that would satisfy your requirements.
2.  Next, I'll look to see if a reliable source out there has already 
published a backported package for the OS repository.  This would be a 
package designed to work within the constraints and package availability of 
your version of CentOS.  I know I've used Dag Wieers (sp?) repositories 
before for RPM-based systems in the past with good luck.
3. Third, I'll try a backport myself.  This really depends on the package in 
question and what library requirements it will have.  I download the source 
package from a newer OS version (in your case, the next release of CentOS) 
and I try to compile it for the local system, so I can at least keep my 
repository dependencies usage consistent.
4.  Finally, failing all that, I'll compile something from source and keep it 
as confined to /usr/local as possible.  You could go one step farther if 
you're going to deploy this to multiple servers and create an RPM package 
from your custom compiled installation, but that's up to you I guess.

-N