Rick Kelly wrote:
> 
> matthew green said:
> 
> >i've heard reports that it does work but no reports that it doesn't
> >except the hearsay from david laight (which i've never heard before
> >but it wouldn't *really* surprise me.)

I worked in the development group of a company that was using the sBus
expander to get similar connectivity with a standard motherboard than
they had got with the previous mBus/sBus motherboard that had two
mBus-sBus bridges and hence two completely separate sBus - giving two
onboard SCSI and 10 expansion slots (they only wired one ethernet up
though!).  This was all about getting up to the connectivity level of
the previous vmeBUD based sparc systems.   These systems didn't run
solaris - indeed they don't have openboot so won't run solaris.

Basically the sBus extender we were using caused much grief! 
Investigations into what is had to do (rather than just believing that
it worked) made us realise the impossibility of making one work 100%
reliably.
> 
> The Sun labeled boxes worked fine when I used them with SS5 and SS20
> systems running Solaris 2.3 and 2.5. We used them in an air traffic
> control project at Raytheon and stuffed them with serial cards.
> 
> Other third-party boxes that we tried sucked.

Possibly the sun boxes were working the same way PCI-PCI bridges work -
accept the cycle on the source bus then generate a new cycle on the
target bus (generating cycle reruns on the source bus).  The ones we had
were definitely trying to do it within the main cycle.

It is also possible (I can't remember, its 5 or 6 yars ago) those system
ran a 33M sBus.

> They were doing X.25 (blech) data connections.

Nothing wrong with X.25 - well not any of the X.25 stacks I've written
anyway.
were you using suns x.25 or the one we wrote (I had nothing at all to do
withthat transport stack!)?

	David