Re^2: performance of counting keys in a big hash

[demerphq]

Use the force Luke!

/* hash structure: */
/* This structure must match the beginning of struct xpvmg in sv.h. */
struct xpvhv {
    HV*         xmg_stash;      /* class package */
    union _xmgu xmg_u;
    STRLEN      xhv_keys;       /* total keys, including placeholders 
+*/
    STRLEN      xhv_max;        /* subscript of last element of xhv_ar
+ray */
};

and

/* the number of keys (including any placeholders) */
#define XHvTOTALKEYS(xhv)       ((xhv)->xhv_keys)

/*
 * HvKEYS gets the number of keys that actually exist(), and is provid
+ed
 * for backwards compatibility with old XS code. The core uses HvUSEDK
+EYS
 * (keys, excluding placeholders) and HvTOTALKEYS (including placehold
+ers)
 */
#define HvKEYS(hv)              HvUSEDKEYS(hv)
#define HvUSEDKEYS(hv)          (HvTOTALKEYS(hv) - HvPLACEHOLDERS_get(
+hv))
#define HvTOTALKEYS(hv)         XHvTOTALKEYS((XPVHV*)  SvANY(hv))
#define HvPLACEHOLDERS(hv)      (*Perl_hv_placeholders_p(aTHX_ MUTABLE
+_HV(hv)))
#define HvPLACEHOLDERS_get(hv)  (SvMAGIC(hv) ? Perl_hv_placeholders_ge
+t(aTHX_ (const HV *)hv) : 0)

HvKEYS(hv) is the thing that implements scalar keys %hash

[hdb]

I think I will have to carry you around the swamps of some unknown planet for much longer before I will ready to dive into Perl's source itself to answer questions in this community...

[demerphq]

:-)

Seriously tho, its not as hard as people think. In fact a lot of the time its easy. All of the code for perl's hash implementation lives in three files, hv.h, hv_func.h, hv.c.

hv_func.h is the home of the new hash functions if you want to look at the options.

[space_monk]

Dammit Jim, I'm a doctor, not a C programmer! :-) _{ok it's a little cross genre, but what the hell}