Re: run.c translator

[Thomas Bushnell, BSG]

Marcus Brinkmann <marcus@gnu.org> writes:

> We are getting into muddy water again, because I was unprecise. 
> If we call the application A, the translator T and the forked program P,
> then I have the normal file interface open(), read(), write() between A
> and T, and a bidirectional pipe created with pipe() between T and P.
> Note that I will replace this with two pipes, so I can close one without
> closing the other.  I have created the bidirectional pipe with pipe().
> Standard pipes are bidirectional on the Hurd (at least I could read and
> write in both directions, I tested it!).

I'm pretty sure that's a bug; standard pipes shouldn't be
bidirectional.  Roland, what do you think?

> What I was asking about is the interface between A and T.

Ok, so what exactly *is* between A and T, if it's not a pipe?