Python's garbage collection was Re: Python reliability

[Paul Rubin]

Steve Holden <steve at holdenweb.com> writes:
> Until someone does the experiment this stuff is bound to be
> speculation (what's that saying about "premature optimization"?). 

40 years of practical Lisp implementation efforts and around the globe
and hundreds of published papers on the subject might not be directly
Python-specific, but they're not what I'd call a total vacuum of
experimental results.

> But I can foresee that there'd be problems at the outer edges of the
> language: for example, sys.maxint would have to be reduced, and this
> in turn would lead to reduction in, for example, the theoretical
> maximum length of sequences.

if we're talking about 1 tag bit, sys.maxint would be 2**30-1 at the
lowest, which means the objects in the sequence would have to be
smaller than 4 bytes each if more than sys.maxint of them are supposed
to fit in a 32-bit address space.  Since we're using 4-byte pointers,
that can't happen.  We may have a worse problem by running out of
virtual address space if we use a copying GC.  Of course, on a 64-bit
cpu, this all becomes irrelevant.