Knee-Deep in SLIC

Immortal Wombat · February 1, 2003, 20:15

I have a question.

Whilst testing out some of the deeper workings of the BattleAftermath event (damn I'm cool), I messaged myself all the values that the event called.

The following results were from a battle where the attacking army lost.

So for BattleAftermath, we have from the docu:

Code:

BattleAftermath(army_t, location_t, unit_t, unit_t, int_t, int_t, int_t)

So not having a string for army, I passed in the army[0].size in, and unit[0].type and unit[1].type, the location as {.x},{.y} and the player numbers.
The attached picture is the result.

army[0].size = 97970400 ?? erm... hows that then?
location coords are correct.
units: 1st one is a mystery even to the game apparently, but it brings up a message I've never seen before... (would this be the type of the unit in the attacking army that got wiped out?)
2nd one comes out as 87 (which was the .type of one of the defenders)
player[0] comes out fine - me
player[1] doesn't work at all
and the int at the end is 1. Maybe this means "attackers lost" and 0 is "attackers won"

But the army size thing is a mystery...

Pedrunn · February 2, 2003, 02:29

Quote:

Originally posted by Immortal Wombat
But the army size thing is a mystery...

I have seen this bug with other event. I have tested the ArmyClicked(army_t).
As i figured out you cant know the size of another civ unit in string form.

In my testing with the army[0].size revealed in a message. If you own the unit it gives the correct number but if the owner is another cive big crazy numbers show up like the one you got.

Martin Gühmann · February 2, 2003, 10:40

Ben could you give us the exact string that you used in the message box I think you forgot something concerning player[1].

-Martin

Peter Triggs · February 2, 2003, 11:03

Weird stuff. I tried:

Code:

int_t armysize;
int_t armyowner;

HandleEvent(BattleAftermath) 'BattleAftermathTestpre' pre {

    setstring(0,ID_pre);
    armysize=army[0].size;
    armyowner=army[0].owner;
    message(1,'BattleAftermathTestM');
}

HandleEvent(BattleAftermath) 'BattleAftermathTestpost' post {

    setstring(0,ID_post);
    armysize=army[0].size;
    armyowner=army[0].owner;
    message(1,'BattleAftermathTestM');
}

messagebox 'BattleAftermathTestM' {
show();
Text (ID_BattleAftermathTestTxt );
}

with strings:

Code:

BattleAftermathTestTxt "{action[0].text}: armysize=
{armysize}, armyowner={armyowner}, 0={player[0].civ_name_plural}, 
1={player[1].civ_name_plural},
\n unit0={unit[0].type}, unit1={unit[1].type}"

pre "pre"
post "post"

The resulting data is the same whether reported pre or post and just depends on who won the battle.

army[0] is the attacking army: if it loses the battle, it ceases to exist. It looks like unit[0] is some attacking unit and unit[1] is some defending unit. Depending on who won the battle, one of these will be the invalid unit and we get that strange error message.

Immortal Wombat · February 2, 2003, 14:43

Quote:

Originally posted by Martin Gühmann
Ben could you give us the exact string that you used in the message box I think you forgot something concerning player[1].

Oh yeah, looks like I screwed up the opening brace {

I can't give you the exact string, I removed it when I used the tester message for something else.

What I wondered was whether since army[0] does not exist (a pain in the arse btw), army[0].size returns the address of the army size?

Dale · February 2, 2003, 16:47

That large number could be the memory address location of the army. A lot of programmers use this shortcut so they don't need to pass the entire class data into new functions, as it causes multiple copies to be in memory.

The Big Mc · February 5, 2003, 05:43

why where you testing this out is there a mod on its way.

Martin Gühmann · February 5, 2003, 13:16

Quote:

Originally posted by The Big Mc
why where you testing this out is there a mod on its way.

I don't know if Ben used it for a new mod possible it is some LOTR stuff or something else. The main information of this thread is how you can test event handlers check the values of the build in variables that are usefull for debugging and that you should extract information from invalid objects. Yeah this will give you sometimes DebugSlic=Yes errors.

So testing this stuff has the purpose of getting information and to satify our curiousity. To get ideas for new stuff.

-Martin

Immortal Wombat · February 5, 2003, 14:37

Battle tracking in the LOTR scenario.

The Big Mc · February 6, 2003, 07:02

I would like to see it when it is done

MrBaggins · February 6, 2003, 08:52

Out of curiousity, do we know for a fact that *this*

Quote:

Quote by MrOgre
Important note! As this documenation is being prepared, the first patch for CTP2 is about to be released. Unfortunately a bug with functions was discovered too late to fix. Specifically, in some cases, using members of unit, army, city, and location variables that are function parameters may not always work as expected. There is, however, a workaround. Copy the function parameter to a local variable and use that variable instead.

was fixed, in every case?

The Big Mc · February 6, 2003, 09:08

Could be a lazy activsion thing. But you have a cat in hells chance of getting anything out of the puddle of eternal wisdom(activsion) I remember the war which was designated as a partial victory as the activsion staff got very offensive and would not talk to me.

MrBaggins · February 6, 2003, 09:13

ahh... I'm not saying that there is any chance of knowing for sure... or getting it fixed if it is... but just maybe its a good idea to copy non int arguments to a local var, just in case... so that we don't get unforseen bugs?

The Big Mc · February 6, 2003, 09:32

We could try that.

I would love to know what enemies I am attacking for my hero unit.

MrBaggins · February 6, 2003, 09:42

I still think you'll have the problem of a non-existent (due to the fact that the designers delete the army/unit after it ceases to exist) defeated army...

The best way of dealing with this is knowing what is being attacked before the battle happens....

using, for instance...

Battle(army_t, location_t)
An army attacks a location

The Big Mc · February 6, 2003, 10:01

yes that code could work

Immortal Wombat · February 6, 2003, 11:48

Quote:

Originally posted by The Big Mc
I would like to see it when it is done

Patience, little grasshopper

I should be done this month unless any bugs show up in testing that passed me by so far.

The Big Mc · February 6, 2003, 11:55

hoow wise master I have travelled many nights to try to find a slicer with my skill could you point me in the direction of a master slicer.

Martin Gühmann · February 6, 2003, 14:46

Quote:

Originally posted by MrBaggins
Out of curiousity, do we know for a fact that *this*

Quote:

Quote by MrOgre
Important note! As this documenation is being prepared, the first patch for CTP2 is about to be released. Unfortunately a bug with functions was discovered too late to fix. Specifically, in some cases, using members of unit, army, city, and location variables that are function parameters may not always work as expected. There is, however, a workaround. Copy the function parameter to a local variable and use that variable instead.

was fixed, in every case?

In that case I would recomment to read the small sentence under the code:

Quote:

The author apologizes for this and promises that if there is another patch it will be fixed. But the above workaround should always work.

Most important here is the little word: if. They promissed to fix it if there will be another patch but as we all know there was no more patch therefore this bug isn't fixed.

-Martin

MrBaggins · February 6, 2003, 15:01

weren't there two patches?

1.1

and

1.11

Anyway... I guess its wise to be wary of the warning, and copy these non-ints locally, when you use them.

The Big Mc · February 6, 2003, 15:28

tipical for activsion.

Martin Gühmann · February 6, 2003, 16:00

Quote:

Originally posted by MrBaggins
weren't there two patches?

1.1

and

1.11

Anyway... I guess its wise to be wary of the warning, and copy these non-ints locally, when you use them.

No there is just one patch v1.1 and v.1.11 are idential the only different is the installer. The installer of v.1.1 had some problems on non US/UK English OS versions IIRC. So the content is the same.

To the promisse again they said if and not when, so it was at this time already questionable if there would be another patch.

-Martin

Martin Gühmann · February 6, 2003, 16:09

Wired I should be the last person posted in this thread, but obviously the post wasn't registrated.

-Martin

Martin Gühmann · February 7, 2003, 13:56

Well I just tried this and good a lot of sytax errors:

Code:

HandleEvent(SendGood)'MG_GiveGoodASound'post{
	message(1,'MGGoodMessage');
}

MessageBox'MGGoodMessage'{
	Text(ID_MG_GOOD_MESSAGE);
}

MG_GOOD_MESSAGE "Ressurce hat: {ressourceDB(value[0])}, {ressourceDB(value[0]).Sound}, {ressourceDB(value[0]).SpriteID}, {ressourceDB(value[0]).Gold}, {ressourceDB(value[0]).Food}, {ressourceDB(value[0]).Production}, {ressourceDB(value[0]).Probability}"

The errors were like this:

arguments not implemented
Function calls not implemented

Well I belive this as most of the fields in the Good.txt don't do anything. But this means the RessourceDB function is useless.

-Martin

Martin Gühmann · February 7, 2003, 14:02

And again a not registered post.

-Martin

Immortal Wombat · February 10, 2003, 06:11

I think it just means you can't call functions from the string files. You need to assign it to a global in the SLIC, and then pass that to the message box.
ResourceDB() is an int_f function that returns a number given a string, so it can't be used in the string files.

The Big Mc · February 10, 2003, 13:27

I had a problem at the week end to I put in some slic for my science victory and got an error. I tried to pass a player into a variable and then use that variable to find out the number of cities the player had.