Many thanks!
Tim.
On Fri, May 03, 2002 at 09:12:10PM +0100, Dave Mitchell wrote:
> This patch fixes the memory leak in
> local $tied[n]
> and
> local $tied{foo}
>
> The problem was that the newly created undef SV that is normally stored as
> the new value of the element doesn't actually get stored in the array/hash
> when there's tying going on. So it doesn't get freed at the end of the
> block.
>
> Dave.
>
> --- scope.c- Fri May 3 20:27:11 2002
> +++ scope.c Fri May 3 20:47:10 2002
> @@ -206,6 +206,12 @@ S_save_scalar_at(pTHX_ SV **sptr)
> PL_localizing = 1;
> SvSETMAGIC(sv);
> PL_localizing = 0;
> + /* If we're localizing a tied array/hash element, this new sv
> + * won't actually be stored in the array/hash - so it won't get
> + * reaped when the localize ends. Ensure it gets reaped by
> + * mortifying it instead. DAPM */
> + if (SvTIED_mg(sv, PERL_MAGIC_tiedelem))
> + sv_2mortal(sv);
> }
> return sv;
> }
