Re: hashtable breakage

[<don...@is...>]

 > An alternative to this single bit is to split the hash table into two hash
 > tables, one for the keys that are instances of with-sticky-hash-code and
 > one for the keys which aren't. After GC you rehash only the second one.
 > And any puthash/gethash operations starts out by determining which of the
 > two sub-tables to access.

I do like that much better.

What did you think of my suggestion for use of object "age" ?  Of
course, this assumes a model of incremental GC where the set of
objects of some type that are moved at a given GC is exactly the set
created since some previous GC, and that it's easy to determine the
creation time (as in the previous GC) from the (address of the)
object.  I've lost track of how many different GC algorithms are used
in clisp and which of them might have that property.

Also, is there any hope of an answer to my question about the test?
Even the answer that there's no answer other than reading the code
is a little better than no answer.