Text 324, 86 rader
Skriven 2005-07-05 20:40:30 av Michiel Broek (2:280/2802)
     Kommentar till en text av Maurice Kinal (1:340/401.1)
Ärende: Re: ZLIB compression
============================
Hi Maurice,

Maurice Kinal wrote to Michiel Broek:
MK> Hey Michiel!
MK> 
MK> Jul 04 20:22 05, Michiel Broek wrote to Maurice Kinal:
MK> 
MK>  MK>> For sure.  I need a new antenna.  Care to make an investment?  ;-)
MK> 
MK>  MB> That still couldn't get me in range.
MK> 
MK> Depends on what kind of antenna I add here.  If I can bounce the signal 
MK> off a few satellites we may be able to resolve that problem.  :-)

That will do fine, a small antenna will be enough then.

MK>  MK>> I like bash scripting and if I need anything more substantial then 
MK>  MK>> perl works quite well.
MK> 
MK>  MB> Sure.
MK> 
MK> For sure.  Tried, tested and true.  However, like binkd, it needs work.  
MK> Nothing that MBSE can't handle ... I think.  Almost there.  I'd like to 
MK> put it on the inet connected box as the official BBS there.  Would you be 
MK> willing to support this evil, demented plan?

Why not. 

MK>  MB> When I was doing the hppa port, the whole thing became stable after 2 

MK>  MB> times
MK>  MB> complete building the system. Once you get there carefull upgrading 
MK>  MB> is the
MK>  MB> only way as soon as glibc and buildutils are involved.
MK> 
MK> Right.  I did manage to get a gcc4/glibc-cvs thingy going using the last 
MK> toolchain as the first stage build but then it was totally out of sync 
MK> with too many of the networking apps I use here.  However I may have been 

That sounds familiar. It's glibc that will almost force a recompile of
everything if it's compiled without backward binary compatibility.

MK> able to get away with it by using a prior gcc release as the optional 
MK> compiler instead of the other way around.  However, the math libs although 

MK> successfully compiled failed some critical tests involving floats and 
MK> memory allocation.  Not good.  So I decided there and then to stick with 
MK> gcc-3.4.4 as the default compiler (c and c++) and glibc-2.3.5 as the base 
MK> libs/headers.  Also using linux-libc-headers-2.6.11.2 and 2.6 kernels as 
MK> part of the scheme.  Works quite well amazingly enough but it took some 
MK> patching and jumping through quite a few hoops to get this far.  Even 
MK> managed to get X to behave itself!!!  Unreal.

It is all possible but it can be a real pain to get everything stable again.

MK>  MB> It behaves itself as long as you be nice to mbse. It can bite back.
MK> 
MK> I noticed that the first time and only time I tried it.  I obviously 
MK> needed to take a harder look at it and see what workarounds I might come 
MK> up with so that it'll cooperate with the scheme here, wired and unwired 
MK> LANs as well as the inet.  No ppp here, or POTs for that matter, so 
MK> ethernets are mission critical.

I'm running inet only, if mbse doesn't see serial ports in the config it
behaves as ip only mailer. Very old version were not so good with that, but
recent versions behave very well on the net. It's also flexible with
nodelist entries so it will know how to reach other internet systems without
manual intervention. Only if the nodelist entry of the system to call is
totally screwed it won't work. But a lot of nets have fixed that these days.

And for a private network like you may need, just create a small nodelist
and you be fine. I use it for testing on the lan, a small net with hosts and
hubs.




    Greetings, Michiel Broek

Email:   mbse@mbse.dds.nl
Fidonet: Michiel Broek at 2:280/2802

... For every action, there is an equal and opposite criticism.

--- MBSE BBS v0.71.3 (GNU/Linux-i386)
 * Origin: MBSE Linux BBS. Made in the Netherlands (2:280/2802)