We have an interesting knife edge to walk here. We'd like to
automagically test example code with a minimum of work on the POD
author's part, and yet avoid false negatives. We also have to keep
POD simple to write. On top of that, we'd like to keep it backwards
compatible.
So we've got a few approaches..
1) Simply test all C<>. This requires the least work on the author's part,
and is 100% backwards compatible, but generates the most false negatives.
2) =for example isn't too much extra work (and it is optional), should give
the least false negatives, but its busts compatibility.
3) =begin preamble/=end preamble/paragraph to test is compatible, but its
kind of annoying to write (or even explain) and will run into problems
with multi-paragraph examples. However, it is compatible and should
also generate the least false negatives.
4) =also for example is an interesting possibility. Its no worse to write
than =for example. It reads right. Its 100% backwards compatible.
And it should also generate the least false negatives.
The more I think about it, the more I like =also. It can be made
generic with a definition like this...
=also follows the form: =also <tag> <tag arguments>. It
instructs the POD parser to consider the following paragraph
as both literal text AND as if it was prefix with the tag and
arguments. This allows text which is both displayed on all
POD viewers and yet has special meaning to certain viewers.
=begin also <tag> <tag arguments>/=end also works similarly
for arbitrary blocks.
Barrie, you suggested it, so its your babe now! You have the option
to sketch out a new RFC and start working on this or punt the
responsibility to somebody else.
--
Michael G Schwern http://www.pobox.com/~schwern/ [EMAIL PROTECTED]
Just Another Stupid Consultant Perl6 Kwalitee Ashuranse
<GuRuThuG> make a channel called Perl, and infest it with joking and
fun....it doesnt make alot of sense.
