WISHLIST: overload fallback method should infer from conversion operators

[p5pRT]

Migrated from rt.perl.org#103508 (status was 'resolved')

Searchable as RT103508$

[p5pRT]

From laurent.dami@free.fr

When overloading the stringification operator '""', some string
operations like regex match, concatenation or split () become
immediately available, but some other operations generate an error
(for example string comparison), because Perl requires the 'cmp'
operator to be also overloaded. So ($overloaded_obj =~ /^foo$/) is OK, but
($overloaded_obj eq 'foo') is an error!

The same thing happens with overloaded numification
(operator "0+") : using it for example as an array index works fine,
but performing an arithmetic operation doesn't work because Perl requires
operators '+', '-', etc. to be overloaded
(see L<overload/Minimal set of overloaded operations>).

SUGGESTION : if an overloaded conversion operator is available, then
overload should automatically use that conversion by default for
implementing fallback methods for 'cmp', '<=>', '+', etc.

I tried to look at gv.c to see how such a suggestion could be
implemented, but this is a bit too arcane for me, so I'm sorry,
I'm not able to directly propose a patch.

Best regards, Laurent Dami

[p5pRT]

From zefram@fysh.org

laurent dami wrote​:

SUGGESTION : if an overloaded conversion operator is available, then
overload should automatically use that conversion by default for
implementing fallback methods for 'cmp', '<=>', '+', etc.

The declaration "use overload fallback => 1;" enables such behaviour.

-zefram

[p5pRT]

The RT System itself - Status changed from 'new' to 'open'

[p5pRT]

From laurent.dami@free.fr

Le 14.11.2011 16​:13, Zefram via RT a écrit :

laurent dami wrote​:

SUGGESTION : if an overloaded conversion operator is available, then
overload should automatically use that conversion by default for
implementing fallback methods for 'cmp', '<=>', '+', etc.
The declaration "use overload fallback => 1;" enables such behaviour.

-zefram

Thanks; I didn't understand that from reading the doc

[p5pRT]

@cpansprout - Status changed from 'open' to 'resolved'