in reply to Re: if only blocks were avoidable...
in thread if only blocks were avoidable...

OK, it turns out that this isn't what I need, because if I use BEGIN, I can't use parameters passed to sub in order to decide on the condition. I can only use environment variables, perl variables and other stuff that already exists before anything happens.

What I need is this: 

sub import {
    $_[1] eq 'StoneAge' and use bytes globally; #no such thing.
}

[download]

BEGIN can't achieve that.

Replies are listed 'Best First'.
Re: Re: Re: if only blocks were avoidable...
by liz (Monsignor) on Sep 22, 2003 at 13:56 UTC
    Ok, I guess I missed that particular point. Well, I think it is possible, but it's rather yucky and it suffers from the same problems that AutoLoader and the likes have. Basically, you compile your code in a two step process. First to compile the "import" routine. Then, when the import routine is called, the rest of the source is compiled Consider module Foo: 
    package Foo;
require bytes; # make sure the module is loaded without (un)import

$bytes = 0;    # flag: whether bytes should be enforced
$evalled = 0;  # flag: whether rest of source has been compiled

sub import {
    unless ($evalled++) {               # we haven't evalled the rest 
+before
        $bytes = ($_[1] eq "usebytes"); # should bytes be enforced?
        local $/;                       # enable slurp mode
        eval <DATA>;                    # compile rest of source (afte
+r __DATA__)
        die $@ if $@;                   # die now if an error occurred
    }
};
1;                              # in order for require of Foo.pm to be
+ successful
__DATA__                        # actual source of module starts here
BEGIN {bytes->import if $bytes} # activate bytes if flag set
sub new { bless {},shift }      # example routine

    [download]
    Then, in an example program, you would do: 
    use Foo 'usebytes'; # or something else to not use bytes
my $foo = Foo->new;

    [download]
    Hope this will help you in your quest.

    Liz

[reply]
[d/l]
[select]
      Yes, that's exactly what I need. And it also shows a very interesting feature of perl - I never thought about evaling <DATA>, so ++. Also I always think of eval BLOCK while there's also an eval EXPR that I always forget.
      Thanks.
[reply]

        Note that a single script can't safely use your module twice. This means that any modules that try to use your module can't be safely used with any other such modules. I don't know what your module does, but it seems quite unfortunate to me.

        I'd probably have three packages. Here is a simplistic example: 

        package Your::Module::Bytes;
use bytes;
do $INC{"Your/Module.pm"} . ".pl";
1;

        [download]
        and 
        package Your::Module::Unicode;
no bytes;
do $INC{"Your/Module.pm"} . ".pl";
1;

        [download]
        and 
        package Your::Module;
sub import {
    # require the desired submodule
    # dispatch import to that submodule
}

        [download]

        Note, I'd have to test to make sure that 'use bytes' affects the code included via do, but I believe that it does.

        I'm also a bit curious what particular effect of 'use bytes' you need. I thought that such was mostly of use for the earliest versions of Perl Unicode support.

                        - tye
[reply]
[d/l]
[select]

In Section Seekers of Perl Wisdom