Linas Vepštas

atoi is very standard and very portable. Try to avoid using boost if you can, it is an endless source of headaches.

Hmm. Well, "happy" is not a verb, so it would not have a tense. "to be" is the verb, "are" is present tense. Techically, "are happy" makes "happy" a predicative...

Oh, I see, yes, if PredicateNode "happy" really means "to be happy", then yes, it should have gotten a tense marker on it. I'm reopening the bug, since I figure...

It's not old code. It is code that is in active use, right now, and Rohit is working on it.

The early stages of R2L almost surely should not use TimeNode and SpecificEntityNode -- those only would get generated during the reference-resolution step, when you figure out if all the...

Just to make "programming" easy, I think it would be best to just invent new types: TenseNode, CountNode, etc. Then you could write ``` EvaluationLink CountNode "definite" ConceptNode "John@444" ```...

This overlaps with issue #2048

Amen, I updated 2 -- problem is microphone is turned off while she's talking. Yes, there should be some vocalization state aas part of self-model

The gaussian is almost trivial. I beleive the following will work (I have not tested) ``` (define (gaussian m d) (+ m (* d (random:normal)))) (cog-execute! (ExecutionOutputLink (GroundedSchemaNode "scm:gaussian") (ListLink...

Try again: ``` (define (gaussian m d) (+ (string->number (cog-name m)) (* (string->number (cog-name d)) (random:normal)))) ```