Maybe you could port it from Gentoo ¿?

.-.-.

Or just doing it from scratch? How hard could it be to code a preemptive multitasking kernel for a C64?

Seriously, of all the weird and promising OS attempts I've seen on the C64 thru the years I believe the only ones who actually (kind of) made it, except GEOS and some other obscure dinosaurs, were Lunix and Contiki and probably Wings?

I could easily see why Contiki never gathered much support. 5k of RAM on a stock C64 doesn't get you very far. Lunix on the other hand, I'm kind of surprised it never cought on. Some ppl started coding tools and drivers for it but it quickly faded... I'm sure today, after 10 years of development, it could have been pretty grand. Oh well... it's not too late

i have seriously thought about it, maybe one day.... =)

as for lunix.... i think it was too obscure back when it was released. and development pretty much stopped because poldi completely abandoned it - and as i said before, noone except him does exactly understand how it - or even the development tools - work.

imho to give it a kickstart again two things are needed. a) clean out all the fucked up patches for c128 and atari support. ie, remove the ifdef hell (if such support is wanted it can be readded again properly later). and b) move the whole thing to an activly supported development toolkit (cc65 suite would do really well). actually i once started to move it to ca65 syntax, but gave up after a while, since i couldnt be arsed to find out how some of the "magic" done by luna and friends exactly works and how to make it work with ld65.

_________________
http://hitmen.eu

Yup. Starting over is probably not such a bad idea. After all there are far better dev tools available today where most of the bugs have been ironed out already. Also it could be better adapted for today's needs and expectations, regarding both hard- and software. Making it an opensource project would allow ppl to participate and let it live on in case of Spontanteous Coder Fatigue

well it IS opensource you know =D

_________________
http://hitmen.eu

Hi all.

I'm neither a programmer or an OS guru but to me (a layman) this looks like a similair project, more recently updated and for all I know with sourcecode that is a well documented joy to read (I don't read code so I wouldn't know)

http://picoos.sourceforge.net/index.html

and from the downloads page: "Download contribution package 1.3.2. The package contains add-ons to pico]OS, currently this is networking support code (adaption code for the lwIP TCP/IP stack, HTTP and FTP server code, and demo programs, one multitasking demo for the Commodore 64)"

Hope this is an interesting heads up and not a waste of your time.

B.

Definitly looks interesting and very suitable for the c64. The fact that there are already ports for c64 emphazises this. There remains quite a bit of work to make it into a full fledged OS though but this surely looks like a better way to go than with Lunix.

The network examples are only for Win32 platforms but the lwIP stack should run straight out of the box more or less. Dont know if there are any RR-net drivers written already but there's a device driver skeleton to start out with so it shouldn't be a big deal. My guess is that the code could be pretty much copy and pasted from the uIP driver.

A suitable filesystem would need to be created for starters, then add network stack with RR-net driver and then contributions to the OS could start off with ease, preferably by creating a shell.

Sounds like a piece of cake, doesn't it?

picos sounds good at first, but then again, think.... even contiki (which is much more lightweight even!) uses too much ram and is too slow. lwip alone probably needs half of the available memory in the c64 to work (there is a reason adam used uip for contiki).

_________________
http://hitmen.eu

hmm.. you might be right. I did'nt think that far even

I wonder how much RAM the Contiki GUI eats. Probably quite a lot. If it was replaced by just a CLI, maybe it would be getting more useful.

there you have it ... thats the only reason i would like to have rr-net drivers for lunix ... the CLi
thats just how it should work ... you open one shell for one software and you press f2 loads a new shell if you whant to start another and f3 and so on .. thats it

the gui is cute thugh

[quote]I wonder how much RAM the Contiki GUI eats.[/quote]

surprisingly little actually (if you use the character based one, not the hires gui used in the last "official" build)

_________________
http://hitmen.eu

Or the other way around: How much free RAM would be needed to even make an O/S attempt interesting? I'd say about 30k. That leaves another 30k-ish bytes for the O/S. Is it even possible to do micro kernel, tcp/ip, filesystem and drivers (network) and a console in about 30k?

Console (unbuffered): 1k (screen ram)
StdIO drivers: 4k ?
RR-Net driver: ~2k
TCP/IP: ~10k (uIP)

That leaves about 15k for kernel and filesystem. Tight but possible

just because i am a crazy nut and i always wanted to do this... i have an OS running, almost completely written in C, using a preemptive multitasking scheduler. it has 4 virtual terminals, inter-process communication, pipes, memory management, kernal filesystem access, a proper shell etc ... i am now experimenting with UIP. ... we'll see i doubt it will turn into much more than "look i can do this", but then again i couldnt care less =D (it was a nice learning exercise afterall)

btw right now there are ~20k free memory without UIP loaded. a statically linked app+uip (the examples from uip) can run in this fine. (and the telnet server "hello world" thing already works)

overall the whole thing is much like lunix... just a bit "heavier" .i implemented a bunch of features, like heap allocation owner tracking, which arent really needed for the OS to function, but are very very useful for debugging the whole mess (once you are dealing with multitasking it becomes a lot more "interisting" to debug =P)... i guess all that stuff can be disabled for a "production" build later at some point. (that will also increase free memory quite a bit) and another BIG plus over lunix is, that the whole thing is built with cc65 in mind, so apps can be made very easily and without all the shortcomings involved when making apps for lunix.

_________________
http://hitmen.eu

/me passes the beer to Groepaz

This is awesome! You really are a nut!

And it's all open source I hope? Can't wait to take a look! I'm sure we can squeeze this baby down to have enough free mem to make it useful.

there we go ..... you are a god dude!!!