On Tue, 2009-04-07 at 11:55 +1000, Jeff Waugh wrote: > <quote who="Daniel Pittman"> > > > > Experimenting is fun. Reality is hard. > > > > I notice you omitted the section of my email where, indeed, I note that > > this is from practical experience. > > Sorry, but ROX and GNUstep are experimentations. They don't have users or > vendors or real systems they need to integrate with or previous version > compatibility issues, etc. When I say "reality", I mean products shipping > and an active marketplace around them (which *can* be said for GNOME/KDE). > Then the hairier issues of software support beyond "hey does this stuff > work?" start to bite.
This appears to diminish the experiments that do occur. I can agree with your generalisation however we should not minimise any effort on FOSS, even experiments. What about those scheduling experiments on the kernel, ultimately led to a major performance improvement for me personally. The problems with packaging really are not with the software but with all the prereq management. This is the strength of debian apt, not so much the install code itself. This does not minimise the complexity of the code itself but just where the true effort was in the first place. People often think a program is the solution, often the data feeding it is much more important. Think Redhat before they began merging Debian prereq information, it was really problematic. -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
