John Porter wrote:
> > For more flexibility, the test could look at an inline_tests.t.list
> > file in the cwd to determine *which* .pm's should be tested.
>
> This could be done now, without further ado.
>
> ##<TEST>
> # testing code here...
> ...
> ##</TEST>
>
> Podulation is not needed for that capability.
Agreed. But then, POD is not needed now anyway.
You can get a similar effect with ## :-)
But of course, POD sections are more informative than #s.
IMHO. they would be even more informative if, instead of
using =head2 or =item to document our APIs, we had things
like this:
=method open FILENAME
...
=cut
Or, for you Javadoc fans, even this:
=method
@type class,instance
@usage open FILENAME
@param FILENAME Path to a file on disk
@returns the filehandle on success, undef on failure
=cut
That's why I favor taking generally-useful things and providing
some standardized support for them. Embedding t-tests in
source sounds intriguing, and a standardized POD-based approach
seems like a clean solution.
E'q
