Bug Stomp

If this is not a bug, could someone please explaninit to me? My initial AI includes
def "none" as "tmpwd";

i enjoy (adjnoun) [45,0] {?PF rem (key1) as only "tmpwd"; ?}
~(key1)?
~You like (mem-tmpwd), eh?

"I enjoy rock music" and "I enjoy rock" do not load tmpwd. What's worse, they seem to delete tmpwd entirely. I get the match on (key1)? that is expected, but the second Response always fails, even when it was the only one. The debugger reports that (mem-tmpwd) does not exist, and that is simply NOT true.

I can't tell if (key1) is not getting stored in tmpwd, but there is my default in there if nothing else. One possible explanation is that it is storing a null (key1) in tmpwd, and since tmpwd is now a null, it's dropped.

I found this one. The (key1) is loaded into tmpwd AFTER the Response fires, even though it is lined up with the KP. So once you have loaded "tmpwd", then the PREVIOUS (mem-tmpwd) will fire the second response. You think I'm crazy? Try it.

You have to have the (key1) to store the memory..You like (key1), eh? then after it is stored you can use (mem-tmpwd) ie. You like (key1), eh? Then I will give you (mem-tmpwd)

In other words I don't "have" (key1) at this point:

i enjoy (adjnoun) [45,0] {?PF rem (key1) as only "tmpwd"; ?}

If not what DO I call (adjnoun)? There is clearly something about handling the parts of speech that I'm not getting.

i enjoy (adjnoun) [45,0] <?PF rem (key1) as "tmpwd"; ?>
~(key1). (mem-tmpwd)??

First pass: "I enjoy green gems."
No match.
Second pass: "I enjoy red sunsets."
red sunsets. Green gems??
Third pass: "I enjoy blue days."
blue days. Red sunsets??

It is discarding the default tmpwd and then using the previous one for each match.

The following code is being passed over, even if I feed it the same KP repeatedly. I ask "What color is your bra?" and the ONLY answer I get is "My bra is a {some random color}." That I consider a bug. But what is really troubling me is that this set, when it works, requires me to load (key1) into a tmp variable and then test the variable. I can't, for example, test {?PF if (key1) is "bra"; ?> because the importer rejects those AI Scripts. And I can't set tmpwd for the scope of the set, because if I try {?PF rem (key1) as only "tmpwd";?} on the KP, it either doesn't load key1 into tmpwd at all, or it loads it AFTER selecting a response. (What is truly maddening is that sometimes it actually works as desired in Transcripts, but I have NEVER gotten it to work correctly in the debugger.)

So apparently the only way to do what I'm trying to do is to load a variable with key1 for EACH conditional Response, and then check that variable against the content of key1. I'd say, "That can't be right!" but I'm learning not to think so....

what color * your (typeof-clothing) [100,0]
~My (key1)? Black.
~You ask about my (key1)?
~My (key1)? Excuse me? {?PF rem (key1) as "tmpwd"; if (mem-tmpwd) is "panties|underwear|bra"; emo: -1; ?}
~What color is your (key1), (name)? [-4:1]

You're right, you can't use <?PF if (key1) is "anything"; ?>, you have to save it to a memory. I guess that is a bit of a hassle, if you don't need it again.

Try putting this as your keyphrase:
what color (is|are) your (typeof-clothing) [100,0]
and this for the script:
<?PF remember "(key2)" as only "tmpwd"; ?>

The soft wildcard could throw things off. As for the rest of the keyphase, you'd need aiscript in each response, so the third response only shows up under those certain conditions. Also, in the third response, you need to take out the <?PF rem (key1) as "tmpwd"; ?>, since it'll be taken care of in the keyphrase.

Spikes KPs aree not that broad

Spikebot: You wanna slip away for a minute, luv? I'll let you blow out my candles.
Ruthie: I think you're supposed to get shot with an arrow or something.
Spikebot: You wanna slip away for a minute, luv? I'll let you blow out my candles.
Ruthie: What kind of music do you like?
Spikebot: Those strange people who keep asking me to bite them see to think so.
Ruthie: My bunny poops in the house
Spikebot: Those strange people who keep asking me to bite them see to think so.
Ruthie: I have a jar full of jellybeans in my bedroom
Spikebot: Can I have a jar full of jelly beans in your bedroom?
Ruthie: I don't get it
Spikebot: Can I have a jar full of jelly beans in your bedroom?
Ruthie: You're cool!

I had something similar this week (that I can't find at the moment), where the convo did a weird jump. It went something like:

Other bot: (***)
Sonora: Response for (***)
Other bot: Something else entirely
Sonora: Another response for (***)

So the AI Engine was hiccupping, or something.

If you look at the debug page closely, the AIEngine keeps track of "Your Last Message:" and "My Last Message:" Maybe they're getting erroneously passed into the preprocessing stage?
I've always wondered if we can use any of the keys the AIEngine uses internally, like (submod), (vmod) and (obmod). Anyone tried these in responses?

>> I've always wondered if we can use any of the keys
>> the AIEngine uses internally, like (submod), (vmod)
>> and (obmod). Anyone tried these in responses?
I am using them more successfully than I am the regular stuff like (verb). Check out the PF2 News for an update on how they parse the sentence.

Warning: The data contained in (submod) (vmod) (obmod) does not add up to an actual quote, because the sentence is parsed into root forms and rearranged, as you can see in the debugger. The various pieces are VERY useful, though!

Mick

good heavens - I've been here for nearly a year, and I've only just found the "PF2 New Features" page! I'd checked the "View Past News" back to the year dot early on, but I always skip merrily past the News page on my way in, with just a quick glance at any new headlines.