Re^2: Constant names come into conflict with Perl style?

All caps subs are reserved

Unfortunately, at some point last year the Perl 5 Porters accidentally broke this rule. There was a problem with cloning in threaded environments, and a magic subroutine named CLONE_SKIP was ushered in to deal with it.

Despite all the eyes of the people who worked on it, it escaped into the wild and it wasn't until a couple of months later that someone noticed that the name violated the guidelines (in that it really should have been called CLONESKIP).

Such is life.

• another intruder with the mooring in the heart of the Perl

Comment on Re^2: Constant names come into conflict with Perl style?

Replies are listed 'Best First'.

Re^3: Constant names come into conflict with Perl style?
by sgt (Deacon) on Feb 09, 2007 at 13:37 UTC

Thanks for pointing this out. I actually remember the discussion on p5p and it was interesting, see the CLONE_SKIP story.

So the underscore is not so good actually :(. The next best thing could be a simple "postfix".

use constant MYPACKAGE_c  => 'Universe::Tiny';
use constant RWXu         => 0700;
use constant MYLOGGERc    => \do{local *FH; *FH};
[download]

Actually a postfix like "c" or "_c" (or even "const" if you don't feel lazy) is good *visually* to remember they are gotchas like remembering when {}ing is actually necessary! (stressed by PBP)

MYPACKAGE_c->factory();
print {LOGGERc} "important login data";

#!/usr/bin/perl

use strict;
use warnings;
use constant LOGGERc => \do{local *FH; *FH}; # silence warnings

open  LOGGERc, ">mylog" or die; # gotcha
print {LOGGERc}  "info...", "\n" or die;
print {\*STDOUT} qx(cat mylog);
[download]

decent_shell% perl -MLWP::UserAgent -Mconstant=PKGc,LWP::UserAgent -e 
+'print +PKGc->new();'
[download]

[reply]
[d/l]
[select]


Pathologically Eclectic Rubbish Lister
	PerlMonks