XSB Src (SVN)-old

Added case in incr_eval_builtin for PSC_SET_INTERN, which was missing.
This seems to work, but Terry should check to see if it's how he
intended it.

Improved end-of-file detection in cases of unicode, but still not perfect;
the unicode for 0xffffffff, if there is one, will be considered as eof.
At least it's better than all unicodes that look like negative numbers
indicating eof, which is how it was...

Changed po answer subsumption to allow nonvariable terms in the
aggregate position, and to treat the variables in that term as existential.
Pointed out by Vijay Saraswat.

Changed how comment was done to quiet gcc compiler warning.

* Fixed several introspection routines so that they expand heap correctly when using load_solution_trie and friends
* Eliminated some more dead code from incremental tabling routines.
* Upped the default heap/local stack size and margin by 4 for sequential only. This hadnt been done in quite a while, so it is time.
Could probably bump it up even more.

Took out all references to affected_gl and changed_gl, which supported eager incremental tabling.
This dead code elimination probably wont make anything perceptably faster, but will make debugging easier,
and will make extension to multi-threading just a little easier.

Commented out some dead code.

Fixed return_lazy_calllist (a critical piece of code for incremental tabling) to properly determine the amount of heap size needed
for returning a subgoal. The problem was that the size determination that had been used assumed that no registers were otherwise used, and
overwrote the registers used in this builtin (which is called by an interrupt handler for the purposes of recomputing invalid subgoals).

There are a number of similar cases that need to be checked in other routines that are not so heavily used. I'll be fixing those in the next
few days.

Fixed bug in return_lazy_affected when doing view maintenance.

Better heap allocation checking for returning tabled subgoals -- I hope1