Re: Help compiling oskit-mach

[B. Douglas Hilton]

Farid Hajji wrote:

> [Discussion moved to address@hidden
>  
Ok. I'm now subscribed. My apologies for the delay in this response,
I've been learning the proper care and feeding of Volvo automobiles
as my summer wanes and the snow clouds build somewhere over the
horizon :-)

> > I've followed the L4/Hurd effort for several years now, casually checking in
> > on their websites every so often. Not much so far. It really sucks that I
> >    
>
> which websites are your referring to?
Oh just the GNU site, http://www.gnu.org/software/hurd/l4-hurd.html
I think there is a Savannah site similar to sourceforge too, but lost track
of it (or could be confused with gnumach/hurd)

> > What about the "Fiasco" L4 variant. I would have thought that it would
> > be rather ideal for Hurd in that it is written in a much more portable 
> > manner
> > than L4 itself. Hurd aims to be the ultimate cross platform environment, so
> > getting extra ports for free right at the start is a really good thing.
> >    
>
> Hurd/L4 will use L4ka::Pistachio,
>  http://l4ka.org/projects/pistachio/
>
> Fiasco implements an older L4 API/ABI, which we don't plan to use
> in Hurd/L4. Only Pistachio currently implements V4 API.
Yes, after revisiting the L4 site, I see that Pistachio is also written 
in a more portable fashion
effecting most of the benefits that Fiasco claimed, but none of its 
disadvantages.

> CPU speed is not the issue here. The Hurd is currently being
> redesigned to better fit L4 (and future uKernels) semantics.
> L4 is actively developed and supported by the research community,
> unlike Mach
Right. Having personally messed around with gnuMach, OSKit-Mach, etc, I 
would not hesitate
to say that Mach has grown into a big scary monster. Its inner workings 
are tough to grasp, and
it is written in an archaic style which is difficult if not impossible 
to work with. I'm not claiming
that L4 is necessarily any better (even though it probably is) just 
saying that Mach is a real
mess when it comes to making it work on modern hardware. The best thing 
that could be done
for Mach is rip out 90% of the code, rewrite that remaining 10% from 
scratch, and start over.
This would be about where L4 was several years ago, so there is a 
definate advantage to going
with the current up-and-coming ukernel just for your stated reasons.

Since I cannot effectively boot Mach anymore on my newish A7N8X/XP2400+ 
box, I figure I'll
try and pitch in with you L4-Hurd folks. I'm kind of swamped at the 
moment with a whole lot
of family-type stuff to attend (yeah, I'm almost an old fart now), but 
once the snow flys around
here I typically hunker down and work on some GNUish development projects.

Will be monitoring the list for a while, and I'll kick in with some 
debugging logs and other
miscellanea as time permits.

Cheers!
- Doug