Bug Stomp

The preprocessing doesn't work correctly for all numbers greater than 100,000.

I found this in a debug session:

Message: 'hundred sixty thousand two hundred three' Time: 0.02
Message: (spell-corrected) 'hundred sixty thousand two hundred three' Time: 0.02
Message: (preprocessed) '60303' Time: 0.03

Same with million.

It's looking for a number in front of the first hundred, and million, too.

I spent the weekend updating and reworking certain aspects of the AI Engine. Along the way I found and fixed several issues, including the issue with the expressions coming up angry. Let me know if that seems to be fixed, or if any other issues crop up. I put some extra debugging lines for expressions in Debug mode.

I seem to have a case of xnoneits, and no facial expressions at all, with the flash face.

Guest1145: would u no who wrote the bible
Bildgesmythe: So, how are you doing today..I am almost perfect

Yet in debug the bot works

You: would u no who wrote the bible
Bot: The Bible was written by many men in many countries over thousands of years. It is the word of God.

When the bots start chats with each other they only use endless x-initiate statements like "Hi, How are you?" "Hi Wanna chat?" "Welcome, nice to see you again ..." Their entire conversation is made up with statements like that. I always end up killing their talk when the box pops up ...out of mercy.

(Sometimes they say "I have too many gotos in a row" too in a bot-to-bot chat.)

Seeks don't work; they run into BLAB.

xmemory is triggered sometimes even if no memory is set. (ok, you could say, then give your bot a default response for this case, which is probably the best advice here.) (And bots with the default xmemory response seem to say "I remember you. You're." without filling in the (mem-youare).)

Number parsing still doesn't work. "three hundred one thousand" is transformed to "1300". (Maybe it would be better to deactivate this feature until it is fixed.)

In the transcripts, the names of the chat partners are missing.

I have a plugin of Mayan gods in one of my replies, and decided to add their names seperately in the keyphrases with appropriate explanation. For some reason, one of the names, "yum kaax", always triggers the xnone.

XXXX: What else does it have to show me?
Madame von Scam: Not very much. Oh, maybe this. You will meet a interior decorator with a cueball who will reveal himself to be a descendant of Yum Kaax.
XXX: Kinda farfetched. Who is Yum Kaax?
Madame von Scam: *stirs her cup and arrogantly ignores you*

I tested without the capitals as well, and also tried to insert the key phrase in raw mode.

I'd advise to try a debug, but the debug feature seems to be broken too. Only about the first third of the page loads; it stops at

Found Stored Response Object

PreSerialize: 0.05
PostSerialize: 0.05
Total Time: 0.05


-------------------------------------

chrono: hour has a problem with time spans where the beginning and the end have the same hour number.

(I've inserted spaces after the colons between hours and numbers to prevent character seqences to be transformed into smileys.)

<?PF chrono: hour(23: 00-23: 59); ?>

is rejected with the message:

ERROR: The second number in the hour span should be more than the first. Use 24-hour time notation. 4PM would be 16: 00.

I can use (22: 59-24: 59) in this case, but what about (19: 15-19: 45)?

Keyphrases starting with a caret (^) (used in many Regular Expressions) are still not shown in the language center. (In a search, the caret is sorted in after the letters.)

After a search, the last clicked link to a keyphrase group is still inactive.

RomulusKesher:
I've given my bots a test keyphrase for spelling correction; the spelling correction and preprocessing transforms 'Yum Kaax' to 'Kaax'. Either use this in the keyphrase, or use a Regular Expression.

(A Regular Expression might look like this:

(who|what).*yum kaax(re)

)

Don't forget to use the raw with the regular expression. If all else fails try
[y][u][m] [k][a][a][x] (re) [45,0] <?PF raw ?>

What Mome Rath said - look in Debug for how it is spell corrected, i was able to make Sublimunse respond to a few sentences in Dutch by taking the spell corrected gibberish and using these as the KPs.

I keep getting a weird "fatal error" message. It started when I checked my bots debugging thing.

That's what I'm getting as well.