Re: pipes & ptys

[itschere@techfak.uni-bielefeld.de]

Juergen Lock,

>  well if you dup2() the new device to the programs' /dev/aux...  but
> since Bcon* still is single-char IO programs that use it will still
> be big CPU time eaters so you might want to get rid of them anyway. :-)

THAT will obviously be the biggest problem, I think, although not for me :-)
since I doubt I've got any of those... I think that's called "compatibility",
must have heard that somewhere :-)

Well, bcon*()'s would then have to be converted to fwrite(1,...) calls to
the new device driver. Hmmm, that' not the way it's done at the moment,
isn't it?

> >  ... So you won't need an extra system call, but just a new device
> > driver for the biosfs...
> 
>  hmm looks i was not clear enough again :)  that was exactly my point 4...

Yes, looks like I didn't understand it. In my opinion with the above mentioned
changes you won't need it AND are still backwards compatible.

bye,
TeSche