DavidW

On the original bug: see Nathan82's thread for details, but it originates in a bug in the Dark Side of the Sword Coast's priest-spells component. That bug in turn confuses SCS and generates the problem. The simplest workaround is to uninstall that component of DSotSC. Since that's at the very beginning of your mod install order, you might instead want to insert this line into another mod's tp2 (SCS, or one that installs before SCS) if you know how: STRING_SET 34735 "Selune's Blessing" Alternately, use Near Infinity to set string 34735 to 'Selune's Blessing'.

Got it. It's a straight incompatibility between SCS and the 'DSotSC priest spells' component of DSotSC; no other mods needed. (And it breaks pretty much all anti-magic spells, not just breach.) Here's what's going on, for anyone curious. SCS dynamically collects a list of all player-usable spells that Breach can dispel. It assumes that the existing description for Breach contains a possibly-incomplete list of those spells. The name of each spell is replaced in the description text by 'DW_PLACEHOLDER'. The comma-separated sequence of DW_PLACEHOLDERs is then compressed into a single DW_PLACEHOLDER. And finally, that DW_PLACEHOLDER is replaced by the newly-generated list of all the spells Breach actually takes down. It's obviously not resilient against arbitrary changes of Breach's description, but it works for Breach as presented in the original game, in the Enhanced Edition, and in SR. DSotSC introduces a new protective spell, Selune's Blessing, that is taken down by Breach. BUT it doesn't give that spell a name: instead it hardcodes the 'name' field of the spell to '34735'. I don't know why, as that string isn't 'Selune's Blessing', or anything similar, on any of the IE games. (The string 'Selune's Blessing' appears only in IWD2, and in a different slot). So this is a bug in DSotSC: the spell is misnamed. Unfortunately, on BG2 (and so on BG2EE, BGT, and EET) string 34735 is blank. And that means SCS's name-substitution goes mad: it replaces every blank space in the spell description with DW_PLACEHOLDER. Contraction of that mess down to one DW_PLACEHOLDER fails, and so hundreds of DW_PLACEHOLDERs get replaced with the whole 'spells affected include' string. So: on the one hand, this is caused by the DSotSC bug: if the spell were named properly, things would work fine. On the other hand, I should have learned by now to write code that is more robust against things being broken by other mods: it was rash to do a substitution on an unknown string without checking it was sensible. I'll fix this next time I release SCS (and since this is a critical bug, albeit only on a specific mod combination, I'll try to do so relatively soon - this weekend, maybe). In the meantime, the simplest workaround is just not to install that component of DSotSC. If you really want it, put the line STRING_SET 34735 "Selune's Blessing" into the tp2 of DSotSC before installing. Thanks to Bartimaeus and Nathan82 for help diagnosing it.

OK: I can reproduce it on DSotSC+SCS 5900. Now to diagnose it...

2: distinguish 'spell' in the engine sense from 'spell in the in-game sense. 3: not true in either 2nd or 3rd edition. (Also, the Xd3+Xd3 thing isn't true on EE - there's a save-for-half flag instead.)

Aha. Thanks, that’s terrific. I’ll see if I can reproduce it myself, and then work out what’s going on!

The breach spell description is probably getting corrupted when SCS 5900 is installed. What I'm interested in is why. I'd very much like to see its description immediately *before* SCS 5900 is installed. Could you uninstall just 5900, then see what it is? I don't just mean 'is it corrupt?', I mean the exact description.

Which items weren’t being taken? And can you post your weidu.log?

IIRC SCS is doing STRING_SET to edit the string, not putting an updated string onto the spell.

So much for that theory.

I assume the NI description is still that weird endless repetition of stuff that you posted a while back. What language are you playing in?

At what point? When version 32 released, about a year ago (or 18 months ago for the playtest version). As for why: there were three design principles driving the redesign of the install system for v32. For instance: - I have long recommended that users install the SCS spell-system changes, because those changes are assumed by the AI and not using them leads to behavioral glitches in enemy spellcasting. So in v32 they are installed by default (but can all be deactivated via the ini, if someone knows that they are going against the recommendation and wants to do it anyway). - I have long advised against giving HLAs to every 18th-level wizard in SoA, but I still used to get complaints that it was unfairly difficult. Now you can still do it, but it's a 'legacy of Bhaal' difficulty, accessible (unless you're playing on LoB) only via the fine-tune-difficulty dialog. I don't think there are any install choices available in v30 that aren't still available in v33, but the default options, given to non-power-users who don't make use of fine-grained customization, are more streamlined and offer less chance of a problematic install. I appreciate your feedback, but it seems to be very much the minority view, so this is unlikely to be reverted. A couple of minor points: - it was never the case that changes as specific as the Guarded Compound staircase locking were called out explicitly in the install process. (That particular change came in v32 and was to avoid various reported glitches, but the lich in the Crooked Crane has been doing it since SCSII was in beta). - I have always had the view that if you don't read the readme, it's your own lookout. That said, the installer's description of 'Initialise AI components' could probably be more explicit that the initialisation includes spellsystem tweaks.

I think that's harmless. No promises though.

I’m not sure I understand the question. Angel’s diagnosis of the incompatibility is correct, but it’s not really something SCS can fix. (My own implementation of the Lanthorn quest doesn’t have this problem, but that’s a quite different Quest in any case.)

Does GiveObjectGoldGlobal not accept a global argument with a negative value?

Cool. OK, that supports my suspicion that CDIBOMBB is bugged somehow. I'll investigate. (When I get a chance...)

It wasn’t trolling, was it? ...or am I missing a joke?

I'm wondering if this is a language-dependent issue. The weirdness originates in the SCS component that dynamically generates descriptions for antimagic spells, and that component relies on search-and-replace moves that aren't always robust on language change. Unfortunately that's very hard for me to sort out, as I don't speak German. I would be interested if it can be reproduced, in a non-English language, on a clean install of SCS.

OK. Try removing the file cdibombb.spl from your override and see if that removes the crashes.

I don't really know. The price aTweaks pays for deciding to be installed after SCS and reliant on its structure is that it would need to be updated any time SCS itself updates. And unfortunately it's no longer maintained, so that's not happened. I can't allow for it on my side and I'm not willing to accept the design constraint of permanently conforming to the assumptions about SCS's structure that aTweaks makes, especially since I don't actually know what they are. Somebody would have to go through its code carefully and see just what assumptions it's making and where.

I don't know for certain. But aTweaks makes quite specific allowances for SCS's fiend-summoning structure, and that structure in turn changed quite drastically when v32 was released. So there is a realistic chance of serious bugs... but I don't know if there actually are any.

Have you ever seen a summoned beetle successfully cast its gas-cloud attack?

I see the case for this. But I don't feel confident it's unequivocally a bug or mistake, which makes me reluctant to change it. (I also think it's good for the occasional creature to be vulnerable to Greater Deathblow - otherwise why take it? - but I agree that the Drakes aren't the ones I'd choose in my own mod.) Maybe I'll compromise and tweak this in SCS's version of Improved Abazigal.

I don't *think* I edit SUMMON_DEATH_KNIGHT (or the Death Knight script) on an SR install. I think I just use the baseline resource.

Hmmm. Yes, that was the bug I had in mind. But I'm puzzled that my detector isn't picking up the issue. Any idea when exactly the crash occurred? Did one of your characters finish a spell? Did an enemy spellcaster?

It’s in the readme.