of tea leaves and half inching

Some blighter has just stolen most of my PCs, so I've lost
my last month's
worth of saved p5p emails, plus any recent in-development
patches.  So if
I don't follow up on any recent threads, this time I have an
excuse 

PS I now know what the fault is in MM_VMS.t, and had a patch
which was
making things far worse!

-- 
There's a traditional definition of a shyster: a lawyer who,
when the law
is against him, pounds on the facts; when the facts are
against him,
pounds on the law; and when both the facts and the law are
against him,
pounds on the table.
    -- Eben Moglen referring to SCO

On 22/12/06, Dave Mitchell <davemiabyn.com> wrote:
> PS I now know what the fault is in MM_VMS.t

and that doesn't fit in the margin of this email ? 

On Fri, Dec 22, 2006 at 01:52:23PM +0100, Rafael
Garcia-Suarez wrote:
> On 22/12/06, Dave Mitchell <davemiabyn.com> wrote:
> >PS I now know what the fault is in MM_VMS.t
> 
> and that doesn't fit in the margin of this email ? 



My 'free ops left on the parser stack after die' change
frees the BEGIN
block even though its been successfully attached to
PL_compcv, and so
gets double freed. The required fix has assorted subtle
complications,
which all boil down to the fact that OPs aren't refcounted,
so its quite
hard to reliably determine whether they've been successfully
cleaned up.

-- 
Monto Blanco... scorchio!

On 12/22/06, Dave Mitchell <davemiabyn.com> wrote:
> On Fri, Dec 22, 2006 at 01:52:23PM +0100, Rafael
Garcia-Suarez wrote:
> > On 22/12/06, Dave Mitchell <davemiabyn.com> wrote:
> > >PS I now know what the fault is in MM_VMS.t
> >
> > and that doesn't fit in the margin of this email ?

>
> 
>
> My 'free ops left on the parser stack after die' change
frees the BEGIN
> block even though its been successfully attached to
PL_compcv, and so
> gets double freed. The required fix has assorted subtle
complications,
> which all boil down to the fact that OPs aren't
refcounted, so its quite
> hard to reliably determine whether they've been
successfully cleaned up.

Is there a technical reason why we dont refcount OPs? Or is
it just
lack of tuits?

cheers,
Yves

-- 
perl -Mre=debug -e "/just|another|perl|hacker/"

On Fri, Dec 22, 2006 at 08:52:30PM +0100, demerphq wrote:
> Is there a technical reason why we dont refcount OPs?
Or is it just
> lack of tuits?

because apart from the problem of ops leaking in failed
evals, there's no
need to for it. Adding refcounting would make ops larger and
slower, and
would require someone to go through the 8K lines of op.c to
ensure that
all the right right counting is done at all the right times.

-- 
I before E. Except when it isn't.