On Tue, Nov 26, 2002 at 04:41:45PM +0200, Riku Voipio wrote: > Something is seriously wrong, if a single bug that affects a single > arch can stop everyone else from forward. We need a way to get packages > that are broken on some platform into the distrubution while the > developers of the arch sort out the problem. Not the way it is happening > currently, that everyone has to wait the platform to fix itself before > updated packages get into distribution.
i have a very different opinion on that. i don't any box that's not x86 but i still like all the ports, because it gets debian a bigger audience which means more testing, fixes and potential developers. and this audience is not only bigger but also wider spread, which means debian is put to much more different uses and therefore bugs that might slip unnoticed are discovered. on top of that it is a good thing to support other architectures to avoid monopolies. a lot of the bugs that show up on one architecture are a bug on x86 as well, but just don't show up. we can for example be pretty sure that our software is 64-bit clean because of things like the alpha port, which means that supporting x86-64 and ia64 is a breeze (which is a nightmare for proprietary software vendors, that never thought their software has to run on anything but a pc). if we want to support these architectures, then we have to support them together with mainline architectures, and not treat them as second-class spinoffs of x86-linux. if we don't release on all architectures at the same time and thus force the developers to fix bugs on less-used arches as well, we will basically give up on supporting them at all. which is something i definitely would not like. on a bit wider perspective, there are many examples where non-mainline uses lead to improvements for all. a very prominent example is the 2.5 kernel, which brings a lot of scalability improvements for small x86 servers done by people who give shit about computers like these. cu robert
