> thank you John, and also Michael's
> reply for using the -i option.
>
> I tried -i and my initial pipe read returned the
> "octave:1>" prompt,
> however, sending commands via my pascal function still
> failed to give me any more output to read.
>
> I am not sure I understood the the experiment outlined in
> your reply.
> I ran the "octave --interactive < fifo > foo.out "
> command and it simply
> gave an empty foo.out file; I did not have chance to run
> the cat command
> from another term and octave already quit.
>
> In the past, I only got succeed using pipes to run a .m
> file from
> command line and retrieve the printed results, but that was
> a
> single execution; but now what I am trying to achieve is to
> interact
> with an persistent session (sort of like a GUI), I am not
> sure if this
> is indeed possible with pipe operations.
>
> I also found some possible issues on the pascal side, I
> posted it
> on the Lazarus forum and I will let you know if I get any
> useful
> feedback from there.
>
> Qianqian
>
>
> John W. Eaton wrote:
> > On 7-May-2009, Qianqian Fang wrote:
> >
> > | What makes me curious is why my stdout reading did
> not give me the
> > | octave prompt, i.e. "octave:1>"? looks like once
> octave get started,
> > | it will start something else, which I can not access
> its stdout via
> > | octave's stdout.
> >
> > No, I think Octave writes the prompt and most output
> to stdout.
> > Probably you want to use the --interactive option when
> talking to
> > Octave over a pipe. If you are on a Unixy
> system, try this
> > experiment:
> >
> > in one terminal window, run
> >
> > mkdir fifo
> > octave --interactive < fifo
> > foo.out
> >
> > and in another, type
> >
> > cat > fifo
> > svd (rand (3))
> > fprintf (stderr, "stderr!\n")
> >
> > and then look at foo.out. It should contain all
> of Octave's output
> > except the "stderr!" message, which should have been
> printed to the
> > terminal where you started Octave.
> >
> > jwe
> >
> >
>