On Mon, Mar 03, 2003 at 04:41:51PM -0700, Collins wrote: > IMHO, b.s. Easy now, no need to offend.
> There is nothing inherently "obtrusive" or difficult to implement about a > rational system that provides a simple but effective directory structure for > multiple versions of major packages. There is no more effort for vendors in > supplying major packages in a /usr/pkg directory, for example, than in > scattering the appropriate files all over the /usr structure. It's just a > minor restructuring of the much beloved RPM specs, and extension to $PATH, > etc. Actually, yes there is. For one, there is backwards compatibility. Programs currently expect to find certain things in certain places. If you change the directory structure, then an older RPM won't work on a newer system. Also, unless everyone changes at precisely the same time (which is guaranteed not to happen), then an RPM for one system won't work on another. On a secondary level, consider that there is a very large number of packages to change. This problem is magnified when you look at the large number of distributions that exist today. Not only them, but also any other vendor of Linux software (e.g. Ximian, IBM, the Kompany). The LSB is only useful if the VAST majority of them make the change. If the changes are so great that they'll break much compatibility, no one is going to want to be the first to change. Thus, change will only occur if they all agree to do it simultaneously. Getting such concensus from such a large number of people is very difficult. Even after that's done, you still have the problem of older packages. No, the smartest approach is to *first* make the LSB as close as possible to whatever everyone does right now. Once almost everyone is in LSB, you can move them, in steps, towards a more optimal system. -- Daniel Carrera Graduate Teaching Assistant. Math Dept. University of Maryland. (301) 405-5137 -- [EMAIL PROTECTED] mailing list
