[cells-devel] Filters, anyone?

[Ken Tilton]

Wow, just tripped over my own code in support of filters, aka synapses. I
think I have to make a change.

I manage to eliminate Synapses as distinct objects by /always/ interpreting
the second of possibly multiple returned values as a "propagate?" indicator.
I thought this would be OK since a slot can hold only one value. But!

   (c? (floor (^length) 2))

...ends up feeding the "propagate?" interpretation, so it turned we all have
been getting unnecessary propagation where we innocently returned multiple
values and the second was non-nil.

My proposed solution is to make the kludge even worse. :) A cell acting in
the role of a synapse/filter will have to return 'cells::propagate-yes,
'cell::propagate-no, or (I think there was a third value). Only those will
be meaningful. Anything else gets thrown away.

Thoughts? Alternatives?

kt
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/cells-devel/attachments/20060529/b8842501/attachment.html>