What are the meanings of the hash keys when calling ObjectSpace.count_objects?

Question

In Ruby 1.9 (YARV) you can get a count of all currently allocated objects like so:

ObjectSpace.count_objects

which returns a hash like

{:TOTAL=>1226560, :FREE=>244204, :T_OBJECT=>26141, :T_CLASS=>9819, :T_MODULE=>1420, :T_FLOAT=>287,
 :T_STRING=>260476, :T_REGEXP=>4081, :T_ARRAY=>72269, :T_HASH=>14923, :T_STRUCT=>4601, :T_BIGNUM=>7,
 :T_FILE=>16, :T_DATA=>54553, :T_MATCH=>5, :T_COMPLEX=>1, :T_RATIONAL=>15, :T_NODE=>524818, 
 :T_ICLASS=>8924}

What is the meaning of these hash keys? Some like T_STRING and T_FILE are obvious. I'm particularly curious about :FREE, :T_ICLASS, :T_DATA, and :T_NODE.

Answer 1

Just a guess: I assume :T_ICLASS counts include classes and :T_NODE could maybe stand for AST nodes.

Here's a full list (unfortunately without comments):

#define T_NONE   RUBY_T_NONE
#define T_NIL    RUBY_T_NIL
#define T_OBJECT RUBY_T_OBJECT
#define T_CLASS  RUBY_T_CLASS
#define T_ICLASS RUBY_T_ICLASS
#define T_MODULE RUBY_T_MODULE
#define T_FLOAT  RUBY_T_FLOAT
#define T_STRING RUBY_T_STRING
#define T_REGEXP RUBY_T_REGEXP
#define T_ARRAY  RUBY_T_ARRAY
#define T_HASH   RUBY_T_HASH
#define T_STRUCT RUBY_T_STRUCT
#define T_BIGNUM RUBY_T_BIGNUM
#define T_FILE   RUBY_T_FILE
#define T_FIXNUM RUBY_T_FIXNUM
#define T_TRUE   RUBY_T_TRUE
#define T_FALSE  RUBY_T_FALSE
#define T_DATA   RUBY_T_DATA
#define T_MATCH  RUBY_T_MATCH
#define T_SYMBOL RUBY_T_SYMBOL
#define T_RATIONAL RUBY_T_RATIONAL
#define T_COMPLEX RUBY_T_COMPLEX
#define T_UNDEF  RUBY_T_UNDEF
#define T_NODE   RUBY_T_NODE
#define T_ZOMBIE RUBY_T_ZOMBIE
#define T_MASK   RUBY_T_MASK

The RUBY_T_xyz enum is defined like this:

enum ruby_value_type {
    RUBY_T_NONE   = 0x00,

    RUBY_T_OBJECT = 0x01,
    RUBY_T_CLASS  = 0x02,
    RUBY_T_MODULE = 0x03,
    RUBY_T_FLOAT  = 0x04,
    RUBY_T_STRING = 0x05,
    RUBY_T_REGEXP = 0x06,
    RUBY_T_ARRAY  = 0x07,
    RUBY_T_HASH   = 0x08,
    RUBY_T_STRUCT = 0x09,
    RUBY_T_BIGNUM = 0x0a,
    RUBY_T_FILE   = 0x0b,
    RUBY_T_DATA   = 0x0c,
    RUBY_T_MATCH  = 0x0d,
    RUBY_T_COMPLEX  = 0x0e,
    RUBY_T_RATIONAL = 0x0f,

    RUBY_T_NIL    = 0x11,
    RUBY_T_TRUE   = 0x12,
    RUBY_T_FALSE  = 0x13,
    RUBY_T_SYMBOL = 0x14,
    RUBY_T_FIXNUM = 0x15,

    RUBY_T_UNDEF  = 0x1b,
    RUBY_T_NODE   = 0x1c,
    RUBY_T_ICLASS = 0x1d,
    RUBY_T_ZOMBIE = 0x1e,

    RUBY_T_MASK   = 0x1f
};

I think most of those are rather obvious. The only ones I can't figure out are T_DATA (see @banister's comment), T_ZOMBIE and T_MASK.

BTW: Note that these are not part of Ruby 1.9. They are part of YARV. They might be totally different on other implementations of Ruby 1.9 or even not exist at all. The documentation clearly states:

The contents of the returned hash is implementation defined. It may be changed in future.

In fact, it isn't even guaranteed that the method itself exists:

This method is not expected to work except C Ruby.

(By which the author presumably means that the method is only guaranteed to work on MRI and YARV.)

Answer 2

You can get more information about the T_DATA category by calling ObjectSpace.count_tdata_objects (described here).

I believe that these are native objects controlled by the VM. Sometimes native extensions can allocate them, as well.