A Zillions Bug Log Out | Topics | Search
Moderators | Register | Edit Profile

Zillions of Games Discussion Forum » Designing Games for Zillions » A Zillions Bug « Previous Next »

Author Message
Karl Scherer (Karl)
New member
Username: Karl

Post Number: 358
Registered: 9-2000
Posted on Wednesday, December 05, 2007 - 5:05 pm:   

A Zillions bug (as at Dec 2007) :

The following describes a Zillions bug which often caused stored games (zsg) of my 'Knots'
games to be unable to run when one tried to load them.
This is the reason why I could not provide solutions for these game up to now.

I have now found out what causes it and I am changing the Knots games to avoid
this Zillions bug.

Error Message:
When I saved my Knots games (zsg), I used to get errors when trying to run the zsg
such as :
'The following move cannot be read : move 51 ...= l-v on g7g8....'

Cause:
This is a Zillions bug and has the following reason:
Sometimes (I am not sure in which cases) Zillions translates the code of two
similar consecutive change-type commands wrong:
g7 (change-type l-v ) north (change-type l-v)
translates into the zsg statement (note the surplus blank between 'l-v' and 'on' !)
= l-v on g7g8
instead of translating it into
= l-v on g7 g8
or
= l-v on g7 = l-v on g8

Edit ZSG:
You can easily edit the zsg to get it to work:
Open the zsg with an editor and replace
= l-v on g7g8
by
= l-v on g7 = l-v on g8
Then run the zsg to see whether there are more problems of this type in the zsg.
Once you have fixed these cases, the zsg runs just fine.

Change Code:
You can eliminate these problems by coding
g7 (change-type l-v) north (create l-v)
or
g7 (create l-v) north (create l-v)

However, in general a fix might not be so easy because 'change-type' works differently from
'create' ! For example, the change-type command keeps all piece-attributes active, while create destroys
the old ones and used the piece attributes from the piece definition.

I am now changing the code of my Knots games so you can store and run any zsg without
the need to edit them.
Greg Schmidt (Gschmidt2)
New member
Username: Gschmidt2

Post Number: 43
Registered: 1-2007
Posted on Wednesday, December 05, 2007 - 6:00 pm:   

Hi Karl,

I found your post to be timely. When Ed Van Zon was developing "Explocus", we found we needed to explore some of the finer points of move generation. The problem you described is also seen by plug-ins. A situation can arise where a plug-in generates moves for Z. Later, Z sends the move back to the plug-in to apply it, but the plug-in doesn't recognize it because Z changed it! (though there are ways around this)

I had not intended on becoming an expert on ZoG's undocumented "move string" syntax, but I'm fast becoming one. Below is an excerpt from one of my messages to Ed.

--- snip ---
[1]
"on" always appears when "change-*" follows a "move" and the "change-*" applies
to the position which precedes the "=". This does not occur when "change-*"
follows a drop or create. So there are two different ways to indicate that
a change-* applies to the same position that precedes the "=". For "move" the
position is repeated (b3). For drop, it is assumed to be repeated due to the
absence of "on"

move: man b2 - b3 = xman on b3c3 c2 ; note the "on" and the repeated position b3

drop: x man b3 = xman c3 ; note that a change-type to b3 is implied presumably because the "on" is missing.

but this happens only when an additional change-* is done that affects a new position.

These are legal:

x man b2 - b3 = xman
x man b3 = xman

[2]
Spaces are sometimes dropped or added (see above example for move) to the move string
obtained from the DLL. This creates problems for Axiom since Axiom expects ZoG to provide
the identical move string that ZoG obtained from Axiom when it told Axiom to generate moves.
If that happens, you will get the infamous "ID=-7" error.
--- snip ---

-- Greg
Keith Carter (Keithc)
New member
Username: Keithc

Post Number: 153
Registered: 8-2000
Posted on Saturday, July 05, 2008 - 11:14 pm:   

I am not sure where to put this but the title of the topic fits and there is not general bug reporting area.

Cannon AI has uncovered a bug. If the game is launched directly from the zrf file the tool bar appears on the selection screen. It is not supposed to. In the case of Cannon AI this gives access to the alternate piece set. Trying to change to that alternate piece set produces a non fatal error that a piece bitmap can’t be loaded (wrong path). The path is correct from the zrf that the selection screen launches. That is a different path from the selection screen which does not use the piece. At the selection screen level an alternate piece set is undefined.

Launching Zillions then opening the .zrf file avoids the bug. The selection screen opens without the toolbar.

--Keith

Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: