Strange piece of code!

[solegaonkar]

Hi Monks! I am trying to debug a legacy perl application where I came across the following code

select((select(ERRFP), $| = 1)[0]);

ERRFP is a file handle created at the beginning of the function. This one fails my imagination! Any idea what is the special use of this kind of a construct? Is it doing anything other than this?

select(ERRFP); $| = 1;

[kcott]

select returns the currently selected filehandle. So, this is doing what you appear to expect:

select(ERRFP), $| = 1

then, within its parentheses, it generates a list:

(last_selected_filename, 1)

Taking the first index of that list ((last_selected_filename, 1)[0]) gives:

last_selected_filename

Finally, this leaves you with:

select(last_selected_filename);

In short, you select ERRFP and make it unbuffered, then reselect the original filehandle. All in one line. Haven't seen this is ages. :-)

Update: I just remembered where I did see this last: using format about a decade ago. See perlform - Format Variables for examples of less uglier ways to do the same thing.

[solegaonkar]

Thank you Ken!

[davido]

It selects ERRFP, sets that handle's autoflush mode to on, and then reselects the original filehandle.

What it's missing is a # Comment.

[solegaonkar]

Dave, Thanks for that. But that is what confuses me.. is there any value added by selecting ERRPF again?

[davido]

It's not selecting ERRPF again, it's reselecting the original filehandle... whatever was current before the first select (before ERRPF). select returns the filehandle that was current before the call.

Read it from the inside out:

1. select ERRPF and remember whatever was selected beforehand.
2. Set autoflush on ERRPF
3. Return to the outer select the first element in the list, which is the filehandle that we remembered in step one
4. select that "remembered" filehandle.

[Jenda]

Others already explained what it does, now for a more readable alternative.

If you do not have time to change the FILEHANDLE to a lexical filehandle (or do not dare to make big changes) you can do this:

use IO::Handle;
ERRFP->autoflush();

[Anonymous Monk]

Well HANDLE->autoflush(1);

[Jenda]

Nope. The parameter is optional and the default is true: ... If no argument is given the previous value is unchanged (except for $io->autoflush will actually turn ON autoflush by default). - from perldoc IO::Handle

[Anonymous Monk]

Always search first

select((select(ERRFP), $| = 1)[0]); -- no results

select select $| = 1 -- bah!

site:perlmonks.org "select( select" FINALLY!

How do I flush/unbuffer an output filehandle? Why must I do this?

Wow, searching can be tough with these improved search engines :)