Re: [Gcl-devel] Re: [Maxima] Re: Lisp vs. Java vs. C++ speed comparison time? [LONG]

[Paul F. Dietz]

Camm Maguire wrote:

> I suppose the canonical way to do this is to make use of the two least
> significant bits in the pointer value on 32bit machines to hold the
> type information, but this seems quite expensive as each pointer
> indirection must be accompanied by a pointer truncation.

The usual way to do this is to lop off the tag bits using offset
addressing modes (add -1 or 3 to the register before loading the word,
for example; this can be done as part of the load instruction on most
architectures.)

In C, you'd say something like this, I suppose:

   *(long *)(((char *) p) + offset)

and hope this bit of nonstandardization was properly interpreted.
As a bonus, if p were not of the right type, and if the machine
trapped on 32 bit word loads that were not 4 byte aligned, then
this would trap argument type errors even compiled at safety 0.

        Paul