Bug Stomp

Sorry if I wasn't clear enough.

The special characters are NOT in the page source.

That affects ONLY the popups, NOT the direct chat pages, and NOT the debugs.

(But it is fixed meanwhile, thank you!)

Thanks for the clarification - I'll look into it.

I have an answer for the hello in debug going to blab - that's actually a FAQ. It only treats hellos as greetings when you haven't chatted with the bot in an hour or two. Otherwise the bot considers the conversation ongoing and sends it to blab. You can see the AI Engine recognizing it by looking for "isHello" I think, in Debug.

I know the FAQ page. What I meant is that on the page that opens directly after clicking on "Debug" there is a message allegedly from me to the bot that is "Hello". And this very first message was "considered BLAB" in that case.

On a second thought, this might also explain why bots don't seem to understand when one answers their "Hello" with a "Hi".

Right, a "Hello" is sent from a user when the Debug page opens, which follows the same xhello/blab rules I mentioned above. I suppose an improvement there would be to leave out the "Hello" and have the bot serve up an xinitiate.

Edit: I've made this change.

In other news, I got the Combined Bot Template to just under 20,000 Keyphrases now from the original 36,000. Still tons of work to do there, and it's inspiring some good ideas while I work.

Thanks, now I see my mistake - I had talked to my bot less than an hour before. Should have thought of this.

I've found something strange about the popup texts that end before a special character:

- Most browsers (both on Windows and Linux) show the mentioned behavior (special character and everything that follows is omitted even from the source)

- Iceweasel works as it should

- Konqueror works like most browsers with the default settings. When I set the character encoding of the main window to UTF-8, it shows the complete text in the popup windows, but strange characters instead of the given one; when I change the character encoding of the popup window by context menu, the special character is shown as itself

The following might therefore be related: When I look in the source of any personalityforge page (normal or popup), I don't find any charset directive there.

I wonder whether this is a bug or whether I simply don't understand the concept of (vp):

"what are you talking about" doesn't match "what are you (vp)", but "what are you telling me" does.

(And "what are you telling me" when tested against "what are you (vp)" fills "you" into (postkey).)

Btw, would be nice to have a pattern that would match (vp) as well as a single word that is a verb form.

That's probably it. I've updated the doctype declaration and the set the charset to ISO-8859-1 which should be the default and seems to be correct for displaying umlauts and the like. Do you see a difference?

Sorry, it is still not working. The charset is still missing in the popup windows.

(And there seem to be two spaces missing in the "big" pages, when I compare them to other websites; Firefox renders that source line in red)

Seems to work now. Thanks, Professor!

When the Forge initiates a bot chat while I am debugging my bot, the chat doesn't open in a popup window but replaces the debug window. (Occurs every time.)

-----

There is still a bug with special characters in popup chats.

The following characters are still not displayed in a popup chat, and suppress anything that follows on that line too:
Œ œ Å Ñ Ò Ó Ô Ö ×

There are no problems in normal chats and Flash chats, but in Debug those characters are replaced by white question marks on black squares in Firefox.

(I've looked at the UTF-8 representation of those characters - their second bytes interpreted as ASCII is something that might be interpreted as punctuation:

… ¼ ½ ‘ ’ “ ” – — )

Btw, I had to enter all those special characters as HTML entities in the edit box of this forum.

There is something I've noticed wrong about all the bots in that their conversation is now choppy, they cannot hold a specific topic nor can they interpret responses as responses, instead they view each user input as part of a new topic of conversation