Re: POD from code?

[Corion]

If I really want to do something like this, I do it the other way around, and create my data (not code) from POD:

my %stuff = map { /^=item * (\w+) => (.*)/ ? $1 => $2 : die "Malformed
+ line: $_" }
            grep /^=item */,
            split /\r?\n/, <<'=cut';

=over 4

=item * foo => ...

=item * bar => ...

=item * baz => ...

=back

=cut

Usually I find that approach to be a bit too much magic, but at least for the definition of configuration file entries, I found it helpfull as it forces me to write the documentation for every possible option together with its introduction into the program.

[JavaFan]

split /\r?\n/

Since 5.10.0, you can write that as:

split /\R/

[LanX]

As a side note:

An alternative to parsing here-docs is putting the POD after __DATA__.

This clean separation between data and code could also facilitate automatic generation of this "Pod-Data"-hybrid.

Or even isolating it into a dedicated module, since 'Text after __DATA__ may be read via the filehandle "PACKNAME::DATA"'

[dgaramond2]

Yup, that's a tiny bit on the syntax abuse side for me too.

I don't mind doing this "manual sync" for test code (and in this case I believe this manual sync is necessary due to its very purpose). But for documentation, I'd like it to be automatic if possible. Because otherwise I'll have to test whether my documentation is up to date/complete.