Lisps



>>>>> "Richard" == Richard Fateman <fateman@cs.berkeley.edu> writes:

    Richard> I guess what we've seen is that the default handling
    Richard> of growing a larger heap space in GCL is not right
    Richard> for Maxima, given certain problems.  And that rebuilding
    Richard> on some new systems causes problems.
    Richard>   1. Do we know that the growing of larger heap space
    Richard> is an inherent problem, or just related to our inability
    Richard> to configure it?
    Richard>   2. Do we know that CMUCL and CLISP are better in
    Richard> this regard?

For CMUCL, the heap space size is fixed at startup to a fairly large
size of 512 MB or so.  Once started, you can't change it, but it can
be changed via a command-line option up to some platform-specific
maximum. 

I don't think CMUCL has any problems on allocating objects from the
heap, as long as you don't exceed the heap size.  CMUCL on x86 has a
generational GC that can be tuned in some ways.  This might help make
maxima run faster.

    Richard> I guess I do not have a specific proposal here if we
    Richard> somehow decide GCL should be maintained, but no one
    Richard> will volunteer to do it!  (Pay someone?)

Diversity is good.  Sorry, I can't volunteer to maintain it.

Ray