Add a link to Matt's blog post

diff --git a/lib/indirect.pm b/lib/indirect.pm
index 96ff670451cfdb73751bf28cb2a6adc0e5339ab4..a5843ab3d86e5cd3b5689d53d6a012473401d7b8 100644 (file)
--- a/lib/indirect.pm
+++ b/lib/indirect.pm
@@ -48,7 +48,9 @@ BEGIN {
 =head1 DESCRIPTION
 
 When enabled (or disabled as some may prefer to say, since you actually turn it on by calling C<no indirect>), this pragma warns about indirect object syntax constructs that may have slipped into your code.
+
 This syntax is now considered harmful, since its parsing has many quirks and its use is error prone (when C<swoosh> isn't defined, C<swoosh $x> actually compiles to C<< $x->swoosh >>).
+In L<http://www.shadowcat.co.uk/blog/matt-s-trout/indirect-but-still-fatal>, Matt S. Trout gives an example of an indirect construct that can cause a particularly bewildering error.
 
 It currently does not warn for core functions (C<print>, C<say>, C<exec> or C<system>).
 This may change in the future, or may be added as optional features that would be enabled by passing options to C<unimport>.