11 Apr
2007
11 Apr
'07
8:11 a.m.
I don't know how hard would it be or whether or not it's a feasible or
a crazy idea... but there it goes in the form of several thoughts.
On Tue, Apr 10, 2007 at 06:56:44PM -0400, Dave McGuire wrote:
To satisfy these requirements, a fork is really the
only option.
Maybe not, if what I'm thinking of could work out. Again, I'm not sure
how feasible or hard it could be, since I'm not a coder myself.
Part of the problem is GCC...It is generally a
decent compiler
(much less offensive since 4.x), but GCC itself is bigger than some
entire operating systems.
I don't think that's a major problem, at least initially. One can
always compile stuff on a faster box and install the binaries on the
older (more classic) ones, I think. Of course, it'd be fantastic to
have a lean, fast C compiler to use there, but I think other things are
even more critical to get to the point of needing the compiler.
"Desktop environments" (garbage like KDE
and Gnome, both of which seem to be actively trying to become slow,
bloated pigs like Windows) are also a major contributor.
I don't see DE's as a problem either. I simply ignore them all
together. I never use them and never will, even when (if) I had a dual
core PIV with 8GB RAM or something like that. I just don't
like/want/need them. :-)
It should be possible to take a snapshot of a
current, modern OS
like NetBSD, strip it down to remove all the bloated crap, and
replace the compiler with something lean, like perhaps pcc or lcc.
It would then be maintained separately WITH DISCIPLINE to keep it
from becoming bloated. It is that discipline that NetBSD has lacked
over the past 5-7 years, and now it's a bloated pig.
Here it's when it (finally) comes my idea... I've been trying to get
some modern Unix variant on my 486 notebook (no CD, PCMCIA NIC) for
quite some time now hithout enough (to my likings) success. It's got
16MB RAM, but, for some reason, it recognises only 12. And, of course,
12MB is not enough to run any modern installer. I've ended up running a
Linux distribution called Small Linux on it, but again, it's an old,
poorly maintained distribution running an old Linux kernel.
Many times I've though... man, if I just had a boot floppy of this *BSD
system which would boot on 8, 4, maybe even 2MB of RAM, and which would
allow me to install a really small base system... I could manage
afterwards to get the programs I need on it once I had this base system
(maybe something like 30, 40, 50MB?).
I used to love NetBSD; I've used it on many
architectures since
v0.9. When Ragge first got the MicroVAX-II support working back in
1994 or os, I was one of the first to boot it. But now it is so
bloated as to be almost unusable on all but the newest, fastest
machines with tons of RAM. It is a real shame. I hate the notion of
forks, and I hate the fragmentation of the BSD community, but I've
believed for a couple of years now that a fork & stripdown of NetBSD
is something that would be of benefit to a great many people...for
use on old machines as well as new.
Well, I think that using an optimised kernel, choosing the right
programs to run and so, many operating systems (included NetBSD or any
other BSD variant, of course) could perform quite well on all the
hardware you refer to. I'm talking about 386 most of the time just to
make an idea of about the power one of the boxes I think of might have;
all my Sun lunch boxes (IPX, IPC, Classic, LX) would do, as well.
So, back to the point. Wouldn't it be simply great to have/build a
boot floppy with an installer which is extremely lean, as to allow one
to install the system with very small amounts of RAM (let say 2-4MB).
An installer which would allow you to install an small, fast, lean (but
current) base system? Later on would be time to think of a replacement
for gcc, or X, or whatever software one wants to use (depending on
what hardware one has, also).
Maintenance could not be too hard, I think, since the system would be
an existing already operating system (let say NetBSD to name one), just
installed on a very cared, optimised for resources way. But still
binary packages or even ports could be used (again carefully selecting
what to use depending on the hardware power).
Well, that's enough for the moment for me to keep building ideas around
that, unless other people feel all this isn't stupid, absurd or silly
enough to just ignore it all together, so that further ideas and
conversation can develop.
I think your English is fine! :-)
Thanks, Dave, sometimes one has the feeling of not being clear enough
'cause of the language limitations, you know. :-)
--
Angel @ Granada, Spain
PGP Public key: http://www.ugr.es/~ama/ama-pgp-key
3EB2 967A 9404 6585 7086 8811 2CEC 2F81 9341 E591
------------------------------------------------------
() ASCII Ribbon Campaign - http://www.asciiribbon.org/
/\ Against all HTML e-mail and proprietary attachments