At Sunday 08 August 2010, Ralf Wildenhues wrote:
> * Stefano Lattarini wrote on Sun, Aug 08, 2010 at 06:06:04PM CEST:
> > At Sunday 08 August 2010, Ralf Wildenhues wrote:
> > > No. You save a fork with
> > >
> > > foo ()
> > > {
> > >
> > > ...
> > > foo_result=bar
> > >
> > > }
> > >
> > > foo ARG...
> > > test "$foo_result" = ...
> >
> > Yes, but how do you get foo_result in our case? ;-)
>
> With sed, of course.
Exactly! So:
foo() { foo_result=`sed ...`; }
foo; bar=$foo_result;
instead of:
foo() { sed ...; }
bar=`foo`
Where is the significant improvement?
> > Or am I missing something?
>
> I think you are. You proposed
> result=`foo ...`
> test "$result" = ...
> which is an extra fork over the above, regardless of what foo does
> internally.
Even considering my latest example above?
I'd like to sort this out, since if I am mistaken it means that I'm
probably misunderstanding something important...
Regards,
Stefano
P.S. I'm delibrately ignoring that we might also want to just
redirect the output of foo() to a file, which means that:
foo() { foo_result=`sed ...`; }
would lead us to extra useless forks *and* to extra coding
indirections, i.e.:
foo; echo "$foo_result" > file
instead of:
foo > file
