Bug Stomp

User: Need white wine? Why? Carbonated apple juice is much MUCH better.
bbbruno: I have, I haven't... what difference does it make?

bbbruno's reply corresponds to the keyphrase "have you".

Thanks for your feedback and attention.

The debugger is useful for stuff like this, although not exactly easy to read!

It looks like "Need white wine?" is being interpreted as "Have you need white wine?" (presumably as in "Have you need of..."). So to catch this you'd need a keyphrase along the lines of

^need (re) [30,0]

Yes, you are right. Thanks.

This Sentence: Need white wine ?
This Phrase: Need white wine
Begin Interj Total Time: 0.48
End Interj Total Time: 0.49
Start Get Word Bases: 0.56
Possible Word Forms: (Ne?)(Nee?)
Base: need white wine
Type: v,n n,a,v n,v
Form: _ _ _
PreDirect 'Have you need white wine' Total Time: 0.83
New Sentence: ' Have you need white wine '

Is anyone else getting nothing but x-nones right now? None of my keyphrases are matching.
I could really, really do without x-noneitis just now

Hasn't hit Sonora yet. *crosses fingers*

Just curious, have the page load times been just totally horrible for everyone else, or is it just me? I have not been absent from this place the last several months because I got sick of my bot or anything, I just haven't been able to really use it between Groundhog day and the rotten load times I usually get with PF.

Every so often I get a day like that. Then I usually just log out and don't bother. But other days everything loads fine. *shrugs*

Phantom Gotos
I keep on getting « Too many gotos in a row » answers. When I trace back the conversation there is not a single goto used to get to the responses given by the bot! Does somebody know where they come from? Could it be that they are taken in account even if the AI does not follow the goto statement by choosing an other response?

I've never had a "phantom goto" problem and I use gotos a lot. Could it be that the keyphrase being triggered isn't the one you wanted? It might be worth typing the trigger phrase into the debugger to see what gets fired.

I also have "too many gotos in a row" results, some of which seem undeserved. I went through Irina Khalidar with a fine-toothed comb recently, making sure that there were no gotos as a direct response to a keyphrase (i.e., NOT as a response to a seek), or as a direct response in any of the special "x" expressions. I still get "I have too many gotos in a row" messages frequently, however.

When I try to get them on the debugger, by saying the same things the guests said, I don't get the error.

Recently I was looking over a transcript in which IK was talking to three guests simultaneously. The frequency of "I have too many gotos in a row" was much greater than usual, and many of them occurred in response to inputs that had noeve given trouble before. I'm hoping this fact will serve as a clue to someone more sophisticated than I.

I've had that error a few times, and it seemed stramge because I don't use a lot of gotos. But it hasn't happened frequently enough to bother me much.

I have noticed that the "too many gotos in a row" occurs after a goto has been initiated and there are no valid responses from the response list. If you have conditional responses based on age, gender, etc., you may trigger a valid response where someone who has not declared those variables would not, hence tripping the "too many gotos". Maybe.
Clear as mud!