Re: (resent) Typeglobs vs. SUPER:: (Hook::LexWrap failure)

On 08/10/2007, Brandon Black <blblackgmail.com> wrote:
> I've attached a patch for what I think is a reasonable
compromise
> approach.  In this patch, when the Foo::SUPER stash is
autovivified, a
> new Foo::SUPER::ISA (with magic) which contains just one
entry for
> "Foo" is created.  Because of the way SUPER
is specially handled in
> the fetchmeth functions, this ISA won't actually get used
for
> lookups, but it serves to tie FOO::SUPER into the cache
management
> code in the right way such that its caches are
invalidated at the

Sounds like something not too expensive, performance-wise.
Thanks,
applied as #32065.

> right times.  It also makes the test case in this
ticket pass.

Maybe we could get a regression test for this, too ? 

On 10/8/07, Rafael Garcia-Suarez <rgarciasuarezgmail.com> wrote:
> On 08/10/2007, Brandon Black <blblackgmail.com> wrote:
> > I've attached a patch for what I think is a
reasonable compromise
> Sounds like something not too expensive,
performance-wise. Thanks,
> applied as #32065.
> > right times.  It also makes the test case in this
ticket pass.
> Maybe we could get a regression test for this, too ?

>

Here's three new tests in t/mro/basic.t.  The last two fail
without 32065.

-- Brandon