Just out of curiosity: Porting to other kernels

Under development: PCMCIA, wireless, etc.
Post Reply
Message
Author
User avatar
cb88
Posts: 1165
Joined: Mon 29 Jan 2007, 03:12
Location: USA
Contact:

Just out of curiosity: Porting to other kernels

#1 Post by cb88 »

Can puppy be ported to diferent kernels like L4 http://en.wikipedia.org/wiki/L4_microkernel_family, Mach,pistachio,Adeos http://home.gna.org/adeos/,other micro/nano kernels?
No, I haven't lost my mind i'm just curious.
Micro kernels seem like they might be easier to maitain eg drivers.
If a fast small mircro kernel existed would or could it become the main kernel for puppy?

What I want in a kernel fast gui on 386 4Mb ram usb standardized drivers never need to port/recompile just keep on using the same ones basically everything linux offers and more. when I say fast gui on 386 I mean that puppy would be bootable and cli apps would be runnable from the gui. NO I AM NOT ASKING TOO MUCH! :wink:

Also this site may help make puppy smaller (less inter-compatability?)
http://www.selenic.com/linux-tiny/index.cgi/LinuxTiny

User avatar
Nathan F
Posts: 1764
Joined: Wed 08 Jun 2005, 14:45
Location: Wadsworth, OH (occasionally home)
Contact:

#2 Post by Nathan F »

If you look at Linux live cd's as compared to, say, FreeBSD based ones, the big difference lies in two technologies. One is unionfs, which is barely supported at all, and the other lies in the fact that bsd doesn't seem to have any kind of compressed filesystem comparable to squashfs or cramfs. This makes BSD based live cd's bigger and slower for the same amount of software, although with a full hard drive installation BSD is in some cases faster than Linux.

I don't have very much experience with other kernels, but my assumption is that you would run into similar problems porting Puppy's unique structure to them. Particularly as relates to the microkernels you mention.

LinuTiny, on the other hand, is a very good and promising idea. I imagine there is a very good chance that the slimmed down Linux kernel would support the modules we need for Puppy. It's primary use however is in embedded devices. The standard Linux kernel has been too large to fit on a floppy for some time, but I think that LinuxTiny does. I can imagine an extremely small console system with just the kernel and busybox, would be all you need for a lot of embedded applications.

Nathan
Bring on the locusts ...

User avatar
Gn2
Posts: 943
Joined: Mon 16 Oct 2006, 05:33
Location: virtual - Veni vidi, nihil est adpulerit

#3 Post by Gn2 »

No - the difference lays in services provided by kernel -I.E. user space versus ABI intercept
This is better understood by just what the kernel does.
http://en.wikipedia.org/wiki/Kernel_(computer_science)

BSD BASICLY uses a monolithic kernel/s - but incorporates own uneasy melange

It is suggested if really interested ~ Google for GNU "HURD" - that concept is still a work in progress (GNU)

Plse see :
http://people.fluidsignal.com/~luferbu/ ... nbaum.html
OR > Google for linux microkernel vs monolithic
(So yes, since you qualified your post - sorry but it is asking too much, Esp in light of present understanding)

It is not a matter of whether Puppy CAN be used with a M/Kernel - it is the changeover required -
ALL default inclusions are affected, involving complete recompiles, which would then
break compatibility to any other monolithic structuring
A lot of work vs little practical gain & who is to do it ?

API/ABI calls are NOT handled in same manner !
Conversely, if practical use of potential is to be used - explore present latency of 2.6_xx kernels

To qualify system " responce speed" vs "power" Plse consider the underlying software, run on peripherals > as "potential"
Whereas actual results are far more dependent on USER l

User avatar
cb88
Posts: 1165
Joined: Mon 29 Jan 2007, 03:12
Location: USA
Contact:

#4 Post by cb88 »

Yeah, I know i'm asking too much. I was sort of being sarcastic. It couldn't hurt to reinvent everything done in computer engineering from the cpu up restandardize everything. I do understand the concepts it is just the puting them into practice that is hard. I wrote a little program that calculates PI how can I create larger data types in c++ I assume this is where the problem lies in my program since it cannot calculate any number past "long double"(larger or smaller than) with out creating a new data type right?I'll edit my code in later.

Code: Select all

# include <iostream>
int HugeInit();
int main() 
{
using std::cout;
using std::cin;
using std::endl;
int pi;
long int loops=0;

int HugeInit ;
{
int pi[40];





}


do{

loops=0;
long int loopcount=0;
long double pi=4;
double divisor=3;
double x=4;
int posneg=0;

cout<<"enter the number of times you want to loop calc pi '0' to end"<<endl;
cin>>loops;

while(loops>loopcount)
{
if(posneg!=1)
{
pi=(pi-(x/divisor));
++posneg;
}
else
{
pi=pi+(x/divisor);
--posneg;
}
divisor=(divisor+2);
++loopcount;

}
cout<< pi<<" pi\n"<<loops<<" loops\n"<<loopcount<<" loopcount\n"<<divisor<<" divisor \n"<<"x="<<x<<endl;

} while (loops!=0);
return 0;

}
enter the number of times you want to loop calc pi '0' to end
10
3.23232 pi
10 loops
10 loopcount
23 divisor
x=4
enter the number of times you want to loop calc pi '0' to end
1000
3.14259 pi
1000 loops
1000 loopcount
2003 divisor
x=4
enter the number of times you want to loop calc pi '0' to end
100000
3.1416 pi
100000 loops
100000 loopcount
200003 divisor
x=4
enter the number of times you want to loop calc pi '0' to end
100000000
3.14159 pi
100000000 loops
100000000 loopcount
2e+008 divisor
x=4
enter the number of times you want to loop calc pi '0' to end
Last edited by cb88 on Sat 28 Apr 2007, 19:17, edited 2 times in total.

User avatar
Gn2
Posts: 943
Joined: Mon 16 Oct 2006, 05:33
Location: virtual - Veni vidi, nihil est adpulerit

#5 Post by Gn2 »

Hmm - most OTOH - find if the concept is known. "practical" use follows?
PRACTICAL is as practical Ceams

Use of C++, to manipulate integers ?

Cut the Pi slice as finely as wished - crumbs still remain.

Perls chained in strings may be of great value ?

In language of choice, NOT so to speak

(Hint, em-boldened cases :wink: enfold THE WEIGH? of sum code)

.

Post Reply