Re: conditionally loading modules

[abigail@foad.org]

On Fri, Aug 25, 2000 at 01:40:32PM -0400, Darxus@chaosreigns.com wrote:
> This is bothering me, and I feel like telling somebody about it.
> 
> I still consider myself relatively new at perl, but in each of the 3
> programs I've written that I felt were worth posting to freshmeat, I've
> used a module that is not part of the default perl install.  Because the
> module is not on all systems, and because the program could still do
> useful things without the module, I want to test for its presence, and
> load it if I can, and deal if I can't.
> 
> Perl does not seem to like this idea very much.  That's what's bothering
> me.

So, why not require the module to be present? If you set the appropriate
line in the Makefile.PL (you do use h2xs, right?) its presence is tested
for during installation. CPAN.pm can even automatically fetch and install
such dependencies.

> Beyond the possibility that I may have just overlooked something, there is
> the fact that when I was looking very actively, I found it very hard to
> find information on this subject.
> 
> Sure, you can do 
> 
> $gnupg = eval "require 'GnuPG.pm';"
> 
> And then check $gnupg before attempting anything gpg related.  I'm fine
> with that.  It's a little odd, and I don't think particularly well
> documented, but somebody (Kyle) showed me how to do it, and I can use it.

I find that quite well documented.

> The problem comes with stuff like Date::Calc.  If you require() it, it
> doesn't import its functions, and it gives you errors saying that its
> functions aren't defined when you try to use them.
> 
> The answer I've come up with is this:
> 
> if (eval "require Date::Calc;")
> {
>   import Date::Calc;
>   $dow = 1;
> } else {
>   print STDERR "Connot find Perl module Date::Calc, not generating Day Of Week stats.\n";
>   $dow = 0;
> }

Why print here? If you use warn(), someone who installs a $SIG {__WARN__}
can do something with it.

> The significant part is the import().  It took reading its section of the
> perlfunc man page quite a few times, and a lot of trial and error to make
> it work.  The fact that perl seems to not bother reporting errors for
> these things didn't help.

Reporting errors for what? Usually, Perl is very very good in reporting
errors. Also, I suggest you use Date::Calc -> import; instead of the
indirect object notation. The indirect object notation is quite evil.

> It seems to me that at some point in writing a program, every coder should
> sit back and think "hmm, what modules that I'm using could this program
> function without ?", and then load them conditionally, and deal with their
> absence.  This should be a common coding practice, and well documented.

But why? This isn't C where 99% of the code written on one platform ain't
going to work on someone elses. Most code runs everywhere. Why would I
have to program around the fact someone might not be bothered to install
a module? Isn't that a small price to pay for a program that you make
freely available?

> I feel like I should mention that the only perl book I've gotten around to
> picking up is the blue camel.

That begs the question, *which* blue camel?


Abigail
**Majordomo list services provided by PANIX <URL:http://www.panix.com>**
**To Unsubscribe, send "unsubscribe phl" to majordomo@lists.pm.org**