Referencing subroutine in not-yet-loaded module

[vsespb]

Is this good practice to write code like this:

my $subref = \&NotYetLoadedModule::mysub();

or it's a bug, and only possible way to get reference to such subroutine is:

my $subref = sub { &NotYetLoadedModule::mysub };

I am asking because after 1st example something is broken, and Class::Load::load_class refuses to load NotYetLoadedModule (this differs from require behaviour and thus I filed a bug report)

[tye]

Being able to take a reference to a sub that has not yet been defined has been a feature in Perl "forever", and it seemed to me to not be an accidental result of some implementation detail. So I see no reason to go with a more complex approach.

[GotToBTru]

The alternatives are .. is this good practice or is it a bug? That's confusing to me. You said it works .. but you are filing a bug report? That's even more confusing.

Referencing a not yet loaded module in order to trigger loading on demand might be a great practice for saving on load time and memory requirements, but I would make sure this is well documented in your code.

[vsespb]

You said it works .. but you are filing a bug report?

What works is references to subroutine in not-loaded packages. Bug report against Class::Load, which refuses to load packages after such referencing (thanks that I still have "require" working).

is this good practice or is it a bug?

Exactly. I think those are more probable alternatives.

Referencing a not yet loaded module in order to trigger loading on demand might be a great practice for saving on load time and memory requirements

I am not asking about such references in general, only about \&NotYetLoadedModule::mysub(); vs sub { &NotYetLoadedModule::mysub };

[Anonymous Monk]

Is this good practice to write or its a bug...

Does it work?

I am asking because after 1st example something is broken, and Class::Load::load_class refuses to load NotYetLoadedModule (this differs from require behaviour and thus I filed a bug report)

Well, Class::Load isn't require

The documentation is pretty clear, it uses "heuristics" to determine if a class is loaded, require doesn't do that, so not a bug in Class::Load behaviour ...

its a documentation bug, it should document the actual heuristics, not make an allusion to the origin :)

When I need to load I use the core Module::Load

[vsespb]

Does it work?

yes. but that does not mean that it's a good practice.

[Anonymous Monk]

The extra parentheses are not necessary (and indeed if you look at this expression in perl -MO=Deparse,-p you'll see they aren't being parsed at all how you expect): my $subref = \&NotYetLoadedModule::mysub; is fine.

Class::Load looks in the symbol table in order to check if a module is "loaded" (this is to allow for inlined packages, that didn't come from a file).

The issue was filed as https://rt.cpan.org/Ticket/Display.html?id=108388 and is not resolvable as this is simply how Class::Load works. Generally you're better off using Module::Runtime::require_module, which looks in %INC, a much more reliable mechanism to tell if a module has loaded.

-ETHER (Karen Etheridge)