Version 5 Bug Reports

[grogerson]

Finished Dragon's Eye. Some tough battles here, and I suspect HoW's modified scripts (calls for help) may be working better than in the original. Once we can play completely through I'll have to play both IWD and the conversion area by area with the same party for both games. The only item of note is a good one - no invisible little girl for the party to attack .

[CamDawg]

Problems from Dragon's Eye, level 3: only the undead lieutenants seem to be problematic. They don't seem to want to attack, just stand there and get turned into dust.

Yep, I can confirm this. They're assigned a script (gnundead) in their general script slot that, once they talk to you, has a block that's always true and it blocks all lower scripts (i.e. their combat script).

Argh, argh, argh. Mapping ChangeCurrentScript from IWD is going to take a lot of manual tweaking. I've just started looking into it and it's already causing this issue, preventing some bombardier beetles from using their cloud attacks, and preventing Angaar from opening the gate post-battle in the barbarian camp.

[DavidW]

Problems from Dragon's Eye, level 3: only the undead lieutenants seem to be problematic. They don't seem to want to attack, just stand there and get turned into dust.

Yep, I can confirm this. They're assigned a script (gnundead) in their general script slot that, once they talk to you, has a block that's always true and it blocks all lower scripts (i.e. their combat script).

Argh, argh, argh. Mapping ChangeCurrentScript from IWD is going to take a lot of manual tweaking. I've just started looking into it and it's already causing this issue, preventing some bombardier beetles from using their cloud attacks, and preventing Angaar from opening the gate post-battle in the barbarian camp.

 

Do have a look at the bit of code in the converter that's trying to handle them. Obviously it's not perfect - hence all these bugs - but you might find it works better to fix it than to start from scratch.

[CamDawg]

Problems from Dragon's Eye, level 3: only the undead lieutenants seem to be problematic. They don't seem to want to attack, just stand there and get turned into dust.

Yep, I can confirm this. They're assigned a script (gnundead) in their general script slot that, once they talk to you, has a block that's always true and it blocks all lower scripts (i.e. their combat script).

Argh, argh, argh. Mapping ChangeCurrentScript from IWD is going to take a lot of manual tweaking. I've just started looking into it and it's already causing this issue, preventing some bombardier beetles from using their cloud attacks, and preventing Angaar from opening the gate post-battle in the barbarian camp.

 

Do have a look at the bit of code in the converter that's trying to handle them. Obviously it's not perfect - hence all these bugs - but you might find it works better to fix it than to start from scratch.

Part of it was a very unlucky bout of spot checking prior to that post--the first three scripts I checked had issues, and it's relatively clear sailing since then.

[CamDawg]

Oh, Presio, you lovable scamp. Always casting cloudkills right next to yourself and casting haste on a point instead of an actor.

[grogerson]

Something for later, now that Kulyok has seen Vaarglan & Co. on entering Lonelywood for the first time. There are seven areas where an adjustment to characters would be a good idea, so they spawn when the game calls for them, not be there but hidden. I've had problems with some (noted in the bugs), but when you can see them, even momentarily, or the game pauses because you "see an enemy," it breaks the immersion.

 

AR1005 - Frightened Child in the fishmonger's house.

 

AR2100 - Orogs, and to a lesser extent the combat Arundel in Kuldahar. I've had him appear in all installs of the game, with the orogs. When the orogs do appear, scripted or unscripted, they also attack the townsfolk at some point (bug report on this).

 

AR2103 - Sheemish in the Kuldahar Smithy.

 

AR9100 - Vaarglan and Company and the assassin Purvis.

 

AR9106/07 - The assassin Purvis.

 

AR9105 - Hailee and Ambere Dunn in their house.

 

These are the areas I'm aware of with problem causing invisibles.

[DavidW]

Something for later, now that Kulyok has seen Vaarglan & Co. on entering Lonelywood for the first time. There are seven areas where an adjustment to characters would be a good idea, so they spawn when the game calls for them, not be there but hidden. I've had problems with some (noted in the bugs), but when you can see them, even momentarily, or the game pauses because you "see an enemy," it breaks the immersion.

 

AR1005 - Frightened Child in the fishmonger's house.

 

AR2100 - Orogs, and to a lesser extent the combat Arundel in Kuldahar. I've had him appear in all installs of the game, with the orogs. When the orogs do appear, scripted or unscripted, they also attack the townsfolk at some point (bug report on this).

 

AR2103 - Sheemish in the Kuldahar Smithy.

 

AR9100 - Vaarglan and Company and the assassin Purvis.

 

AR9106/07 - The assassin Purvis.

 

AR9105 - Hailee and Ambere Dunn in their house.

 

These are the areas I'm aware of with problem causing invisibles.

 

This is something that's likely to improve when the CreatureHidden stuff is recoded, following Cam's discovery of the relevant bit. So I'm disregarding this for now. If that doesn't work, then sure, we might think about something a bit more radical.

[lynx]

Read a few of the last pages. Re summoning delays in summon appearance:

IWD has 7 second rounds, while bg2 has 6. I don't know if your converter adjusts all the durations - if it doesn't this could easily be the culprit (prolonged avatar removal).

[DavidW]

Read a few of the last pages. Re summoning delays in summon appearance:

IWD has 7 second rounds, while bg2 has 6. I don't know if your converter adjusts all the durations - if it doesn't this could easily be the culprit (prolonged avatar removal).

 

We're certainly aware of the fact, and I believe we catch them all. But thanks for the reminder.

[CamDawg]

Yxonomei's cloudkill appears to do nothing. Otherwise, things were dandy through DE and the return to Kuldahar.

[CamDawg]

Mage animate dead lacks a *c.bam, causing it to be a blank square in the spell book. Clicking on it to add to memory causes a CTD.

[CamDawg]

A bug from the original game... if you talk to Hjollder and either decline his quest at the last minute or you don't meet the level requirements, Hjollder_quest gets set to 1 or 2. The Contact Other Plane spell will give you only HoW options if this variable is non-zero; it needs to be checking for >=3.

[CamDawg]

Testing: Created a series of rings with bonuses or penalties to speed. Then equipped greater and greater negative speed rings to a PC until it was immobile to establish the party's base speed (9 in both engines). Then brought in a green-circled creature of each animation type and simply raced it and a PC across the screen. I adjusted the PC's speed via the speed rings until I got a match.

I just realized that I left my testing stuff in fix v3, so if anyone wants to take this for a spin you just need to uncomment it from the tp2.

 

The death animation will look fugly because it doesn't blend like spell effects do.

 

Anyways: .cre carries a cast spell on condition (IIRC opcode 232) which is triggers when taking damage, and plays the death animation. To make sure the death animation is played only when the .cre is dead, externalize opcode 215 (play 3D) into an external .eff file that targets a GENERAL of 3 (DEAD). The engine internally sets dead creatures' GENERAL to 3.

 

-Galactygon

Wouldn't this fail if the WV was killed via spell (i.e. a kill opcode) or do those count as 'hits' too?

 

Dragon's Eye level 4 (hotfix 3 installed):

 

Once the battle started, several doors remained locked. As I recall only the door to Marchon and company should remain so. This kept the summoned creatures from coming after the party. Also, when the bard spoke to Albion the normal dialog options appeared, when the paladin did he had only one option. Nice to see those class-specific lines back... Trolls appear to fall properly now.

OK, got it. The fix I put in here for v3 assumed there was only one door using the d4doorg and d4doort scripts, which is not the case. This is fixed locally.

 

So, HoW with fixpack 3: I wasn't able to complete it, unfortunately, because after going to barbarian camp from gloomfrost(which went well) and seeing the entire wylfdane/seer sequence(which went well) and slaughtering all red barbarians in camp and in the hall I still cannot exit the camp. Hjollder says "good luck, my friends" and says nothing else, and no one opens the gate(the healer guy wanders next to them, but does nothing). Bad.

Same report from grogerson; I'll investigate.

 

Other small things: as reported before, wights on the Burial isle sometimes drop wolfskins;

This is IWD behavior.

 

right after arrival to Lonelywood, some evil party briefly flashes next to you and disappears.

Should be fixed in the next core update.

 

Also after Hjollder disappears from Burial Isle, an invisible party-aligned(green-aligned) creature remains there.

It's cloned from the magic mouth creature; the magic mouth patch is not working properly as the invisibility should be permanent. Fixed this locally.

 

Problems from Dragon's Eye, level 3: only the undead lieutenants seem to be problematic. They don't seem to want to attack, just stand there and get turned into dust.

Yep, I can confirm this. They're assigned a script (gnundead) in their general script slot that, once they talk to you, has a block that's always true and it blocks all lower scripts (i.e. their combat script).

Argh, argh, argh. Mapping ChangeCurrentScript from IWD is going to take a lot of manual tweaking. I've just started looking into it and it's already causing this issue, preventing some bombardier beetles from using their cloud attacks, and preventing Angaar from opening the gate post-battle in the barbarian camp.

 

Do have a look at the bit of code in the converter that's trying to handle them. Obviously it's not perfect - hence all these bugs - but you might find it works better to fix it than to start from scratch.

Part of it was a very unlucky bout of spot checking prior to that post--the first three scripts I checked had issues, and it's relatively clear sailing since then.

Yeah, pretty much ignore this. I found a few places where we need fixes; the rest was me forgetting that NI labels creature scripts incorrectly in area files. The undead lieutenants were buggy, though, and are now fixed.

 

Oh, Presio, you lovable scamp. Always casting cloudkills right next to yourself and casting haste on a point instead of an actor.

Presio now casts Cloudkills at the nearest enemy, provided he's out of its radius. Haste is cast on the nearest ally.

 

Yxonomei's cloudkill appears to do nothing. Otherwise, things were dandy through DE and the return to Kuldahar.

Hmm. I think this may be just be a missed magic resistance message, as everything's scripted correctly.

 

Mage animate dead lacks a *c.bam, causing it to be a blank square in the spell book. Clicking on it to add to memory causes a CTD.

Fixed.

 

A bug from the original game... if you talk to Hjollder and either decline his quest at the last minute or you don't meet the level requirements, Hjollder_quest gets set to 1 or 2. The Contact Other Plane spell will give you only HoW options if this variable is non-zero; it needs to be checking for >=3.

Fixed.

 

The final Icasarcht area had animation layering issues, so it's been added to the existing patch.

 

Rest encounter patch is only patching one rest percentage instead of both.

 

Three White Doves was missing it's 5% chance to destroy outer planar creatures. Granted, there's only one in the game but still.

 

Found and fixed a passable wall in the Severed Hand (ar5003).

[Galactygon]

Wouldn't this fail if the WV was killed via spell (i.e. a kill opcode) or do those count as 'hits' too?

 

The following count as "to hits":

1.) any successful attack roll.

2.) any damage suffered through item or spell

3.) if the creature is hit by a spell with the hostile flag

 

The only problem with #1 (but not #2&3) is with launchers: the shell spell fires off the moment the hit roll is calculated (the projectile leaves the launcher). So you will have to place another condition, which is damage suffered (it's in the list of conditions).

 

-Galactygon

[CamDawg]

Wouldn't this fail if the WV was killed via spell (i.e. a kill opcode) or do those count as 'hits' too?

 

The following count as "to hits":

1.) any successful attack roll.

2.) any damage suffered through item or spell

3.) if the creature is hit by a spell with the hostile flag

 

The only problem with #1 (but not #2&3) is with launchers: the shell spell fires off the moment the hit roll is calculated (the projectile leaves the launcher). So you will have to place another condition, which is damage suffered (it's in the list of conditions).

 

-Galactygon

Cheers.

 

David, I'm increasingly convinced that the evade fixes are misbehaving. Whenever my mage casts ice storm, she double-clutches the casting animation, and continues to do so for a couple of rounds while it runs. It interrupts later spells as well--she also aborted a fireball after casting an ice storm due to this.