On Wed, Jul 13, 2005 at 08:33:12PM -0400, [EMAIL PROTECTED] wrote:
> On Wed, Jul 13, 2005 at 04:29:53PM -0700, Michael G Schwern wrote:
> > I don't see the utility of the current behavior, except maybe it was easier
> > to implement, and IMHO its ripe for mistakes like the one pointed out
> > earlier:
> > use strict;
> > package Foo;
> > our @ISA = qw(Foo);
> > package Bar;
> > @ISA = qw(Baz);
> > print @ISA; # Hmm, everything looks ok.
> > print @Foo::ISA; # SURPRISE!
> > print @Bar::ISA;
> > No warnings issued and strict won't save you.
>
> 1) Fix @ISA, or
What @ISA breakage does the above demonstrate?
> 2) Introduce real package scopes:
We already basically have them.
{
package Foo;
our @ISA = ...
...
}
Folks don't like em. It sucks to have to indent an entire package. Same
reason you'll almost always see:
<html>
<head>
<title>foo</title>
</head>
<body>
<p>This is in the body.
<ul>
<li>only now do we indent
</ul>
</body>
Of course this doesn't stop you from doing:
{
package Foo;
...
}
but nobody seems to do that.
> our() was never a problem for me, and if it is changed, some of my code
> may break. I always prefer lexicals over global variables.
>
> Consider code like:
>
> our($MY_GLOBAL);
>
> ...
> {
> package SOME_OTHER_PACKAGE;
> $MY_GLOBAL = $VAR_FROM_OTHER_PACKAGE;
> ...
> };
The feared Club of Backwards Compatibility! Curses!
FWIW I wouldn't do that because it makes noticing that $MY_GLOBAL is not
actually a variable in SOME_OTHER_PACKAGE difficult to spot.
package Bar;
...a bunch of code...
package Foo;
$Variable = 42; $Foo::Variable or $Bar::Variable?
Is that $Foo::Variable or $Bar::Variable introduced via an earlier our?
You can't tell without examining all the preceding code throwing out part of
the point of lexicals and packages: limiting scope.
It doesn't follow the normal $Global, $lexical conventions. If
SOME_OTHER_PACKAGE really needs to access another package's globals its
better to be explcit about it and say $PACKAGE::MY_GLOBAL.
Unfortunately these arguments appear to be a few years too late. I even tried
going back in my time machine to try and change the behavior or our()
before it was implemented but I found that no matter what I did I was unable
to alter the implementation. :(
http://www.newscientist.com/article.ns?id=dn7535
--
Michael G Schwern [EMAIL PROTECTED] http://www.pobox.com/~schwern
ROCKS FALL! EVERYONE DIES!
http://www.somethingpositive.net/sp05032002.shtml
