[nos-bbs] My response - JNOS not being a race car, having stress problems.

[Maiko Langelaar (ve4klm)]

Steven,

> My apologies if my snitty comments about winlink/linux

Heck, no apologies necessary. I didn't think you were being snitty.

> first you commented that running excessive servers and features in
> NOS slows it down, so you run only the packet switching features.

That was actually an observation, I run alot more than just the
packet switching features.

> But when it comes to mentioning Linux, because it allows unlimited
> memory, all of a sudden a 5 megabyte NOS executable is okay ?

That so called 5 megabyte NOS you see is mostly debug and symbol tables,
because in LINUX the default compile includes debugging code. If you strip
the NOS executable (linux command, 'strip jnos'), you will find the real
NOS executable to be about 700 kilobyte or around that area. If you remove
the -g3 debug option from the makefile, you will then see the true size
of the LINUX executable, which like I said is well under 1 megabyte.

> I question I ask is "Does Jnos REALLY Require so much computing power as to
> justify a 2.6 kernel with 1 gig of ram and a 4 GHZ dual core 64 bit CPU?

Gosh, of course not ! I still use an IBM PC Server 315 (P100), it's MORE
then I will ever need to run my NOS stuff. The 486 (66 Mhz) previous to
that was more than I will ever need. You don't need that computing power.

> I know a lot of NOS computer enthusiasts are running such hardware, but
> it seems to me somewhere along the line of NOS going from DOS to Linux
> some people have become convinced that type of hardware is necessary.

That's all wrong. There is simply no need for such crazy computing power
just to run NOS. But, I am sure you will understand that there are new 
users out there that have just one computer at home, and running alot
more than JNOS. JNOS is not their primary application !!! So obviously
they are not going to go out and by a lame PC to run their 'modern'
applications. I doubt alot of new users are going to run a separate
PC just for JNOS, so they might as well run it on something they just 
recently acquired or ALREADY HAVE. It's certainly not necessary, but
it's the platform they choose for EVERYTHING ELSE they do.

I certainly would not go out and buy the latest state of the art
just to run JNOS on it. But if I want only one computer in my home,
is there anything wrong with running JNOS on it if possible ?

That's the only reason I stay on the linux development, to make sure that
JNOS will CONTINUE TO RUN on the newer platforms. It will still run fine
on the oldest hardware.

Guys, JNOS 2.0 is still JNOS 1.11f - It's just got a few more features in
it, that's all - some may say useless features, whatever - it's still the
same old JNOS 1.11f, with some bug fixes, new stuff that does not have to
be compiled in. Yes, it still compiles in DOS !

> I would dare say that setup was loaded down 200 times more than most
> NOS switches, especially when they are situated behind a hardware
> firewall/router, then further seperated by the TUN to the linux NOS.

Your point ? What's wrong with giving NOS a bit more breathing room,
a bit less "stress" if I may, so that it can better deal with the "user"
applications within it ?

> Point being all of this ran fine on a 386 with one meg. My first DOS NOS
> machine was an 4.7 MHZ 8088 XT, wid 640k, I still own it and I still compile
> 111f using bcc 3.1 for it.

And it still will. You can take JNOS 2.0, compile it with BCC 3.1, run it
on your 8088 XT, and it will work. I realize that I have done very little
DOS support over the last couple of years. 2.0f is coming out very soon,
I will try and make an effort to keep it's size down by default.

> This is not yet more DOS preaching, this is an example of the
> extremely small amounts of computing resouces needed to run Jnos.

You're telling me something I already know Steve. You're preaching
to one of the choir members. I too ran JNOS in DOS years ago, I just
decided to move to an environment that personally gave me more options
to play. That's all !

> The concept that everyone wants to have everything under the hood
> of the same car mainly applies to Linux.

I guess Linux gives them the luxury of doing that :-) I've run a few
things under the hood, the performance for me was never an issue, and
that part actually bothers me a bit. In all of this discussion, I have
been wondering from the first post - what's the big deal here ? I can
not say I've ever been plagued with performance issues of such extent
that I'm going to show great concern about it. My systems have always
performed to my satisfaction. I don't get it actually, but ...

But then you might say, well why then did you give us all those snippets
on NOS not performing well when it has alot of processes. That was all
technical mumbo jumbo, in theory, some concepts, stating of the obvious,
and so on. In reality, I've never been plagued by any of it to such a
degree that I'm going to be really concerned about it.

Anyways, I'll stop now ...

> DOS users having been picking and choosing within config.h for years

Of course. It had to be out of necessity.

> however it seems 5 or 6 meg Linux executables that literally have 
> "Everything" under the hood is the biggest argument for using NOS
> under Linux in the first place.

Don't confuse "EVERYTHING" with executable size. My so called Linux
executable here at home has HARDLY ANYTHING, yet it is still 8 meg in
size, UNTIL I 'strip jnos', and suddenly it is 667 kilo instead. Hey,
what gives ??? Again, it's that big because I compiled it with debug
options, so that if it crashes, I know EXACTLY where the problem is
and then I can go and fix it (most times). If the size bugs you,
then perhaps I should start distributing the makefile without the
debug option. You can easily take it out yourself (remove the -g3
option).

> Jnos as a switch is one tiny segment of the population. It is one
> of many different reasons to run and use NOS. On network
> 105 you find find Jnos that have up to 10 ports, with convers,
> AXIP, ax.25 to internet gates, full functioning BBS (With FBB forwarding).

And I'm sure it works fine.

> This a classic sysop gateway intended for end users and it is the very
> reason all this stuff was added to Jnos in the first place.

Of course.

> Claiming Jnos only performs well as an AXIP AX.25 switch under *NIX
> is just as scary as suggesting it is a horribly broken, undocumented
> project on it's way to first beta.

Nobody said that. They said it performed better, it shined as a switch,
and not just under *UNIX, it was more directed to DOS. In any event, it
was all conceptual talk, it was all *theory*, to perhaps help people
better understand a few things. You know as well as I do that theory
is one thing, reality can sometimes be totally the other way.

Hopefully I haven't dug a hole for myself with any of these comments,
if so I apologize in advance, I'm not an expert in this at all, I just
have a few years of experiences with the software, nothing more.

Maiko Langelaar / VE4KLM