On Wed, Jun 22, 2005 at 11:10:51PM +0100, Dave Mitchell wrote:
> On Tue, Jun 14, 2005 at 10:45:17PM +0100, Dave Mitchell wrote:
> > On Fri, Jun 10, 2005 at 11:04:36AM +0100, Dave Mitchell wrote:
> > > I'm vaguely assuming that for tieing, during localization you would *copy*
> > > the magic; then at restore time you just ditch any magic on the new SV,
> > > and stick the old SV back into the typeglob with its current magic
> > > intact. ie
> >
> > I now have a working patch that does just that
>
> and now applied as change #24942
Two further things related to this.
1) I have added an mg_localize() function, which can be viewed as part of
familily of functions that include mg_copy() and mg_dup(). Now, the mgvtbl
structure happens to have been extended to have function pointers svt_copy
and svt_dup which allow a particular magic to override the standard
mg_copy() and mg_dup(); but to provide bincompat, these functions are only
called if the MGf_COPY or MGf_DUP flags are set. Currently I believe
threads::shared is the only part of the core to provide these hooks.
I propose, by analogy, to provide a similar hook for mg_localize();
although I haven't checked closely; I think my patch currently breaks
local $shared_var;
If I don't hear any complaints, I'll do this in a few days time.
2) Nicholas, my patch is specifically supposed to fix all the $#foo
problems; an earlier email of yours included a test file for this; do you
want me to add it to bleed or leave it to you?
--
"Strange women lying in ponds distributing swords is no basis for a system
of government. Supreme executive power derives from a mandate from the
masses, not from some farcical aquatic ceremony."
-- Dennis - Monty Python and the Holy Grail.
