Originally posted by Harishankar
My idea was to continue having "testing" for the base packages but still have the option of installing specific packages from "unstable" if needed. Is that a dangerous combination? I read that this is possible in the documentation, but I am wary of doing this.
Currently I have reverted all repositories back to testing.
No not really dangerous at all I have been doing that for a couple of years. You need to edit the /etc/apt/apt.conf and /etc/apt/preferences files and have them look like this.
>$ cat /etc/apt/apt.conf
The Default-Release line tells apt you want packages from testing, Show-Versions line shows you where the packages will be coming from when using apt-get and Cache-Limit increases the package cache limit so you can have more than one repository cached.
>$ cat /etc/apt/preferences
Pin: release testing
Pin: release unstable
This tells apt that you want packages from testing over unstable during a normal install/upgrade. If you want to install packages from unstable then you use apt-get install -t unstable package_name
and it will install the package you want. If you get errors saying that the package cannot be install because another package(s) is at a lower version than the unstable one you want needs then add the package(s) to the end of the line until it will install without the errors. You may want to look into using the -s option of apt-get to simulate the command then when every thing installs without errors remove the -s from the line and do the actual install, this allows you to check and make sure nothing you want from your install will get removed accidently because of conflicts. And of course you will need to have lines for both testing and unstable in your sources list like this changing the lines to reflect the mirror you are using.