Petr Stehlik wrote:
Gentoo actually has real utility on our current limited setup. With gentoo you can use use flags to determine which features get built into which programs. All the way down to which modules would get built for which apps. This would help each user determine which modules they want to statically link into python, etc. I agree though, this is a poor solution and I'm working on implementing vm and all of that. It's quite a project though and I'm still learning the nitty gritty details of our hardware and stuff. I too think we should be focused 80% on sparemint and at least 20% on the more useful idea of openembedded.Miro Kropacek píše v Po 03. 11. 2008 v 10:49 +0100:And the last note: do you think using gentoo is the best choice? I mean, surely most of Atari users would be happy to compile (in case they even want to compile anything) stuff via cross compiler so in my eyes something like CLFS (cross linux from scratch) or systems with support for more architectures (opensuse for example)Definitely go for opensuse... I'll go for Debian then. Hopefully somebody else will then come up with Ubuntu etc. I believe we can easily reach the moment when each MiNT user gets their own unique distro. Petr P.S. frankly, why gentoo when there was sparemint? Note that I don't like either of it as I am Debian positive (for good reasons) - I am just real curious why this "distro split" come to our 10-users world...
Imagine if our full system was crossbuilt we could have daily rebuilds of ALL of sparemint. I have a Mac Pro at work that could take care of that by itself probably. This would allow us to actually propogate changes to mintlib, libpng, etc through the system fully.
I can understand why he would desire gentoo, I use it on my dvr server. It's not a complicated mess to maintain so why not. Openembedded takes a lot of time and energy to understand for someone not experienced with it, but trust me it is worth it.
The other option which may even be better is scratchbox which is used on the Linksys NSLU2 NAS devices. This system uses QEMU to run native compilers and applications in situations where required. With this there would precisely 0 special exceptions. That is if there is a 68k qemu target. Of course that is targeted specifically at linux so I still feel openembedded is the best solution until we have virtual memory.
Sparemint is wonderful and RPM is a great system but the lack of ability to cross compile is a real problem these days.
Thanks, Mark