Mike Lambert wrote:
[ Unifying Buffer and PMC ]
> As jason said in another message, Dan has changed his mind from
> yesteryear, and decided that buffers and pmcs should be the same
> structure.
Roadmap
-------
1) Hide Buffer and PMC internals, namely
- buflen
- bufstart
- flags
- cache
inside access macros and #defines and unify common flags (e.g.
BUFFER_live_flag <=> PMC_live_flag) to have the same value.
E.g.
str->flags & BUFFER_constant_FLAG => get_constant_FLAG(str)
pmc->flags |= BUFFER_live_FLAG => set_live_FLAG(pmc)
pmc->flags &= ~BUFFER_live_FLAG => reset_live_FLAG(pmc)
The actual changes in *.c can be done file by file, are equally readable
and don't disrupt functionallity.
2) Change Buffer, PMC to a common structure, adjust above access macros.
This change would not be seen from the outside.
3) Cleanup, use the common base object, wherever appropriate, e.g.
headers, smallobject, dod.
Proposal for a common structure
-------------------------------
The base object is Pobj (Parrot object), it's very similar to a current
Buffer:
typedef struct _Buffer {
void * bufstart;
size_t buflen;
} _Buffer;
typedef union UnionVal {
int int_val;
double num_val;
void* struct_val;
char* string_val;
struct PMC* pmc_val;
_Buffer b;
} UnionVal;
typedef struct Pobj {
UnionVal u;
unsigned int flags;
} Pobj;
typedef struct Buffer {
Pobj o;
} Buffer;
A PMC is derived from this structure and has additional fields like the
VTABLE amd next_for_GC (if we really need this).
typedef struct PMC {
Pobj o;
VTABLE *vtable;
} PMC;
It would be also possible to have an extended PMC type for aggregates:
typedef struct APMC {
Pobj o;
VTABLE *vtable;
void *data;
} APMC;
.... or just use the latter for both.
Comments welcome,
leo
