Allison Randal wrote:
The Exception PMC never had a get_bool vtable function implemented,
but it did previously inherit one from ResizablePMCArray. So, that
test was relying on an implementation artifact.
OK, then that was certainly a wart, agreed.
Under what circumstances should an exception report TRUE? Under the
old implementation, since it was using ResizablePMCArray, it would
report TRUE if any data was stored in the exception. That is, it
always reported TRUE, because 'init' prefilled the first 11 slots in
the exception's array.
In Perl 6, truth is supposed to fall back to whether something is
defined or not. So a try with an error occurring sets $!, making it
defined, so "if $! { ... }" runs the block if there's an error. And if
there's no error, $! is not defined.
Parrot shouldn't necessarily just do whatever Perl 6 needs, though, and
in the end we maybe should in Rakudo have "if $! { ... }" really calling
$!.true or 'prefix:?'($!) - and off the top of my head, I'm not sure
which is right. At the moment we just generate:
find_lex $P17, "$!"
if $P17, if_16
So I think this likely needs a bit more thought from the Rakudo side
before we rush and stick a get_bool into the Exception PMC. Maybe
something to work out next week at YAPC...
Thanks for the clarification from the Parrot side of things!
Jonathan
