Question about redirection

Question about redirection

[malte]

Could some kind soul explain this to me ?
	{ echo stderr >[1=2]; echo stdout } >[2=1] > /dev/null
	stderr
	{ echo stderr >[1=2]; echo stdout } |[2=1] cat > /dev/null
	stdout
	<waiting for input>

Malte.

Re: Question about redirection

[Tim.Goodwin]

>Could some kind soul explain this to me ?
>	{ echo stderr >[1=2]; echo stdout } >[2=1] > /dev/null
>	stderr
>	{ echo stderr >[1=2]; echo stdout } |[2=1] cat > /dev/null
>	stdout
>	<waiting for input>


Try

	{ echo stderr >[1=2]; echo stdout } |[2=0] cat > /dev/null


[ Don't ask how long it took me to realise the mistake :-(. ]

Tim.

Re: Question about redirection

[malte]

I'm sorry for not being precise with my question. What I want to know
is why
	{ echo stderr >[1=2]; echo stdout } >[2=1] > /dev/null
and
	{ echo stderr >[1=2]; echo stdout } |[2=1] cat > /dev/null

although lexically similar, differ in their semantics, while

	{ echo stderr >[1=2]; echo stdout } |[2=1] cat > /dev/null
and
	{ echo stderr >[1=2]; echo stdout } >[2] /dev/null

look quite different, but are semantically equal. I became aware of this when
I was asked about something similar by a beginner. I feel it is hard to
explain why >[n=m] differs from |[n=m].

Malte.

Re: Question about redirection

[Scott Schwartz]

The strangeness is that >[n=m] means "the thing on the right reads n
but gets m", while |[n=m] means "the thing on the right reads m but
gets n.  Backwards!  When I first saw |[n=m] I thought m and n should be
swapped so it would work more like >[n=m], or maybe the other way
around.

Re: Question about redirection

[John Mackin]

    The strangeness is that >[n=m] means "the thing on the right reads n
    but gets m", while |[n=m] means "the thing on the right reads m but
    gets n.  Backwards!

Not at all.  Or at least I don't think so.  All it takes is the right way
of looking at it; I think I have a right way, and I hope it will answer
Malte's question as to how to explain this to an rc novice.

When we write >[2=1], we think of that as "create two by duplicating one",
and view it like an assignment statement (left receives right).  When
we write |[2=1], we think "the left process's two gets connected to the
right process's one", and don't use the assignment metaphor.

I think that as long as we think about this in the right way, there
should be no problem.

OK,
John.

Re: Question about redirection

[Scott Schwartz]

The point is that |[m=n] is confusing because it uses '=' in such a
different way than >[m=n].  Wouldn't |[m|n] have been more obvious?
Aside from the deeper semantic issues, it is "m:=n" vs "m->n", but with
the same symbol '=' stuck in the middle.  Yes, you can "think about it
the right way", but why should one have to?  Grumble, grumble, grumble.  :-)

By the way, does anyone miss "if not" (from Tom Duff's rc)?  I don't.
Anyone really like overloading static to mean both "own" and
"private"?  I don't.