[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[no subject]
- To: (BUG LISPM) at MIT-AI
- Subject:
- From: DLW@MIT-AI
- Date: Wed ,21 Dec 78 10:07:36 EDT
RMS, Moon and I talked for a while about the use of the new data types
for the implementation of streams. The <- function seems to be buying
two things: first, it causes the variable SELF to be bound, and secondly,
it allows messages to be send to such Lisp obejets as numbers and arrays,
as well as objects, entities and the other functional objects. The
former function can simply be put into the microcode for DTP-ENTITY,
and the latter is not needed for any of the applications for which we
will be using "entities" as part of Lisp (they are only there for people
who do NOT be planning to use Lisp, but rather to do everything with
entity-like operations). Therefore we agreed that the microcode should
be changes to bind SELF for dtp-entities, and that after some more time
for them to settle down, streams will be implementable as dtp-entities,
and still invoked in the manner of regular Lisp functions. Existing
code to dfeine streams and call streams will still work, and gradually
all streams can get converted so that changes to the main "stream" class
will not have to be duplicated in STREAM-DEFAULT-HANDLER (the latter being
a clumsy way of implementing superclasses which will eventually be flushed).
We think this scheme should gain us all of the important advantages of
entities without most of the problems.