On Tue, 2006-01-17 at 22:57 -0800, Danny Yoo wrote:
> > > The terminology that the original poster uses ("references to another
> > > object") sounds a lot like the original usage of symbolic "soft"
> > > references in Perl.
> > > (http://www.perl.com/doc/manual/html/pod/perlref.html)
> > >
> > > Perl programmers, for the most part, avoid them now because they're so
> > > error prone. So if the original poster is thinking about symbolic
> > > references, then we should encourage the poster to look into
> > > dictionaries, since dictionaries are a fairly direct replacement for
> > > that usage.
> > >
> > I did not know Perl programmers were moving away from references. That's
> > new to me, and an interesting point.
>
> [Note: I'll write some Perl code below. Probably bad, nonidiomatic Perl
> code. I also have a high chance of saying something wrong. If I do so,
> someone please correct me. *grin*
>
> Also, whenever I say "Perl", I'll mean Perl 5 --- I understand that Perl 6
> has a model that's similar to Python now.]
>
>
> Hi Victor,
>
> No, no, I'd better clarify this to try to avoid someone getting angry at
> me. Maybe I'll dig myself into a deeper hole, but oh well.
>
> Perl has a distinction between "soft" symbolic references and "hard"
> references. I think symbolic references may have been a language mistake;
> I think good Perl style these days has been to phase them out of usage.
>
> Perl programmers seem to be perfectly happy with "hard" references:
> (http://search.cpan.org/~nwclark/perl-5.8.7/pod/perlreftut.pod)
>
>
> Python programmers like myself struggle a little when we talk about this
> because, truthfully, we don't think about this much. In Python, we pass
> references around without with abandon: we might not even realize it!
> Values in Python are all references. For example:
>
> ### Python ###
> >>> msg = list("hello world");
> >>> greeting = msg
> >>> id(msg)
> 466880
> >>> id(greeting)
> 466880
> ##############
>
> In Python, both 'msg' and 'greeting' are names that refer to the same list
> value. We see this because they have the same id(). And if we apply
> mutation on the list value, we'll see this from both names:
>
> ### Python ###
> >>> msg[1] = 'a'
> >>> msg
> ['h', 'a', 'l', 'l', 'o', ' ', 'w', 'o', 'r', 'l', 'd']
> >>> greeting
> ['h', 'a', 'l', 'l', 'o', ' ', 'w', 'o', 'r', 'l', 'd']
> ##############
>
>
>
> Perl makes a distinction between values and references to those values, so
> the following shows different results:
>
> ### Perl #########################
> mumak:~ dyoo$ cat > test.pl
> my @msg = split //, "hello world";
> my @greeting = @msg;
> print [EMAIL PROTECTED], "\n";
> print [EMAIL PROTECTED], "\n";
> $msg[1] = 'a';
> print "@msg\n";
> print "@greeting\n";
>
> mumak:~ dyoo$ perl -w test.pl
> ARRAY(0x180b560)
> ARRAY(0x180b5b4)
> h a l l o w o r l d
> h e l l o w o r l d
> ##################################
>
> Here, we see that @msg and @greeting are really separate things in Perl:
> the values are in different memory locations, and assignment copies the
> data structures around.
>
>
> To get the same aliasing effect as in Python, we'd have to explicitely ask
> Perl to reference and dereference:
>
> ### Perl ###########################
> my $msg = [split //, "hello world"];
> my $greeting = $msg;
> print $msg, "\n";
> print $greeting, "\n";
> $msg->[1] = 'a';
> print "@$msg\n";
> print "@$greeting\n";
> #####################################
>
> So Perl folks have a slightly more complex conceptual model: they make a
> distinction bewteen values and references. Python just has references.
>
> There are tradeoffs here. On the one hand, Python folks get caught off
> guard when they first encounter list aliasing. On the other, Perl folks
> get caught off guard when they first try to make a hash whose values are
> themselves lists. *grin*
>
WOW. Great explanation. Don't worry about a deeper hole.
Your examples once again reminded me of one of reasons I started looking
into Python instead of continuing with Perl. It is not the
functionality, but the convoluted scripts can get. One can hardly
understand the code.
Thanks again.
Victor
>
> Hope this helps!
>
>
_______________________________________________
Tutor maillist - Tutor@python.org
http://mail.python.org/mailman/listinfo/tutor
