Persistence: a proposal

[Patrick Logan]

>>>>> "David" == David Gadbois <gadbois@cyc.com> writes:

    David> [Scott, Kelly, and Patrick discuss Smalltalk 80-style
    David> object tables.  It is kind of weird to think that all of
    David> this stuff, which still seems so cutting-edge, was
    David> originally proposed 20 year ago when I was still in grade
    David> school.

Think of this: Xerox had in-house publishers running personal
workstations with bitmapped displays, WYSIWYG editors, laser printers,
and ethernet in 1974.

    David> Will the tots of today be raving about parametric
    David> polymorphism in the 2010's?  I don't think so.]

Which is why a brand-new OS should be aggressively challenging the
standards of the 80's.

    David> Having object-table handles does make it easier...

One thing to keep in mind is that things are never quite what they
seem. Although the "Smalltalk-80" programming model presents object
tables and Gemstone carries that idea forward conceptually, a lot of
tricks are played to make it really work.

    David> It also does give up half the address space, already a
    David> precious commodity on 32-bit machines.

But, like the Texas Persistent Store from U. Texas (damn, another
reference to "research"), a 64-bit address space can be swizzled into
a 32-bit system. The effective address space doesn't have to be limited
to 32-bits. Especially if the LispOS should run on Alphas, the
upcoming 64-bit Intel/HPs, the 64-bit SPARCS, etc. Then a 64-bit
address space can map into 64-bit systems and 32-bit systems
simultaneously.

Which is another reason starting with RScheme and Texas is an idea to
consider. Although I don't know if Texas is already a 64-bit system or
simply an idea.

Another idea tossed around for 64-bit addresses is to use 64-bit IEEE
floating point as the OID format. All non-float addresses are encoded
within NaN format. Does someone have a reference to what this looks
like. I haven't found anything on the net that describes the actual
layout of the bits for NaN.

-- 
Patrick Logan                 mailto:patrickl@gemstone.com
Voice 503-533-3365            Fax   503-629-8556
Gemstone Systems, Inc         http://www.gemstone.com