Peter Slegg wrote:
This is true, this is one of the more important reasons to do it - but further in some cases performance is improved. Someone mentioned that fork is very resource heavy right now which is why the shell scripts run so slowly right now. Further, many many unix projects dynamically load plugins... pidgin, apache, python, perl, etc etc etc Our python build is ridiculous. On the build time the package maintainer has to try to make a decision as to which modules every single person will need. Inevitably some necessary module will be missing and many modules you don't need will be included. The mess that ensues of having to rebuild python just to run a particular package is the worst. It's the same story for a lot of things.On Fri, 10 Oct 2008 20:07:43 , Miro Kropacek <miro.kropacek@gmail.com> wrote:True but we all aim for VM and shared libraries. At least I know I do. I ordered that BSD 4.4 book as well ;)While I'm in this evil mood (?) -- why we need it after all? I mean yes, it's cool and modern and it allows some nice things like mmap() but from Atari point of view? GEM stuff wont use it, demos wont use it, games wont use it. It will be nice only for ports from Linux world and how many apps / libs you actually *need* to have linked dynamically? Arguments like "when you update libc, all applications will gain from it" isn't right here since number of applications is counting in tens so it's OK to rebuild everything on cross compiler using some build script and release new distro when some serious bug happen to be fixed.Wasn't it said earlier that small number of developers able to build distros (until they get bored) is a problem, so wouldn't having the ability to dynamically link libs would be an advantage ?
The memory requires rather are not a problem. Still if one copy of bash was loaded at all times it would probably make shell scripts load and exit faster since it would just permanently leave a copy in memory.I agree that memory requirements are a problem, in most cases it would be ok to load several copies of the same lib (if this makes thigs easier).
Perhaps. You would want CUPS. Cups if we could port it allows you to direct postscript output to any type of printer. We would need to create/adapt a postscript NVDI driver and all wold probably be pretty easy.Does VM create possibilities to port more *nix stuff like USB and printer drivers ?
Seems like just yesterday ;) I've been very inactive lately but winter is right around the corner and I am mostly caught up on home maintenance.All stuff for the future. We haven't had a FreeMint release in years. Frank, can we have one please ? Peter
Thanks, Mark