On Wed, Mar 15, 2000 at 06:54:07AM -0800, Robert W. Current wrote: >> Linux was once fairly complete at 10M, now it's easily 500M in /usr, >> and nothing in /usr/local or /opt. That's starting to get rediculus. >> The OS is not 500M... That's the issue. So, why is there 500+M of >> crap in /usr/bin?
>Why would it be better to have 500M in /opt? That's the sticking point. >You're saying that you don't want a standard that allows distributions >to put things in /usr, and I'm saying that there are distributions that >want to do things that way. Can you offer a reasoned argument to support >why your way is right and the other way is wrong? Sure, /opt is fine for 500M, so is /usr/local if /opt and /usr/local are unmounted, distroyed, or corrupt, one should expect the OS itself to remain intact. The OS itself should be isolated from what non-essential software does during it's install, removal, and upgrading. Logic seems clear (to me) that keeping essentials in /usr/bin and non-essentials in /usr/local/bin or /opt keeps things cleaner, more logical, and more functional. FHS states that "added software" should not go in "/usr/bin" The issue boils down to this debate to me: If Red Hat Packages it, does that mean it's "part of the base" or is it just "added software" done by a distributer rather than the user. What happens when a user adds more .rpms? This is why I think it should be in /usr/local or /opt, not /usr/bin... Just because it's "a distribution" doesn't mean it's not "additional software beyond the base"
