Re: Issue: HASH-TABLE-GC (no proposal)

[Jeff Dalton <jeff%aiai.edinburgh.ac.uk@NSS.Cs.Ucl.AC.UK>]

> My notes from Fairfax meeting...
> 
> Cleanup meeting:
> 
>  This issue is about the inordinately long discussion on Common-Lisp@SAIL
>  about hash tables.

Much of the discussion was about secondary issues, so it wasn't really
so bad (I think).

>  Masinter said he would try to get Dalton to write up this issue as
>  a proposal.

I still plan to write it up.  I'd wanted to wait until discussion had
pretty much stabilized, and then some other things kept me from getting
to it.

> X3J13 meeting:
> 
>  KMP: There are several kinds of such hash tables. Those that are weak
>       on key link, weak on value, and both.

True.  Weak on key (if it becomes impossible to refer to the key
(modulo the test) the entry can be removed) had the greatest precedent
and seems the most useful.

>  JonL, RWK: Nervous about this because it's "new technology".
> 
>  Gregor: It IS implemented -- just not in Lisp. It is NOT new technology.

Pop11 has this feature as "temporary properties".  Pop implementations
have supported it from some time.  For the purposes of this discussion,
Pop can be considered Lisp with a different syntax.  PopLog Common Lisp
can use the tables in Pop.  Lisp/VM has weak tables.  Interlisp has (I
think) switched back and forth about what hash tables do.  At one point,
they may have been week.  T has weak sets (formerly populations).  T has
weak pointers.

There are a lot of other points.  I think I will be able to summarize
them adequately in the proposal.

Thanks, Kent, for sending the message.

Cheers,
Jeff