Re: device snp build broken?

Hello Jille, Alex, Alexey,

* Jille Timmermans <jille_at_quis.cx> wrote:
> Alex Keda schreef:
> > Jille Timmermans пишет:
> >> Hello,
> >>
> >> snp(4) in -CURRENT is currently not supported.
> >> Ed made a patch for it; which worked for me.
> >> You can get the latest version from:
> >>
> >> http://people.freebsd.org/~ed/mpsafetty/
> >>
> >> (The patch-set also includes other things; like a new /dev/console)
> > 
> > What was the reason not to support the device?
> Ed rewrote the TTY system and left out snp(4):
> * He didn't thought anyone was really needing it
> * Because it reads from a TTY, and writes to another TTY; it would have
> caused problems with blocking (full queues, etc)
> 
> After a while he got convinced to fix snp(4).
> It is now experimental, so only available through his patchset.
> 
> But it will make it into HEAD before 8.0.
> 
> Ed: Please correct me if I'm wrong somewhere

The main problem with snp(4) was that its design didn't really fit into
the model of the new TTY layer that I imported back in August.

The reason the new snp(4) isn't available in SVN yet, is because it
isn't finished. It doesn't return the traditional snp(4) status codes
and doesn't allow descriptors to /dev/snp to switch to a different TTY
after being used.

I really want to fix at least the first issue, but so far I haven't
found {time,interest,etc}. The new snp(4) code is already pretty sweet -
it works a lot better than the old driver. Maybe someone else is
interested in fixing it?

-- 
 Ed Schouten <ed_at_80386.nl>
 WWW: http://80386.nl/