polytope

I think about it this way; level 13+ fighters probably have 3.5 apr (with 2 weapon fighting), assuming you have 3 such fighters in your party (or equivalent) that's 10.5 apr in total (okay, your cleric and thief can contribute in melee, but maybe you only have 2 fighters, or not all of them are using 2 weapon style, so let's assume it balances out). Casting Haste raises each fighters apr to 4 (as it's always a whole number with vanilla haste) or 12 in total, a net increase of 1.5 apr. Casting Improved Haste on one fighter raises apr to 7, a net increase of 3.5 apr... That alone is more than twice as useful as hasting all your fighters in terms of damage (discounting movement rate and attack speed factor for the moment, though they are relevant in many fights). However, regular Haste is not stackable, whereas you can IH each of your fighters to get 21 apr in total, a net increase of 10.5 apr. I suspect Improved Haste and Whirlwind are the main reason so many ToB creatures have way more hp than they should, per 2nd ed.

Yeah, I do feared that not using the original opcode could be a problem, but at the same time it would be the best way to do a "clean job". Regarding the repeating EFFs issue, as far as I remember even if Blade Barrier works through multiple 146 sub-spells you still get 2x rate. I'll try to find some time to test it. Effects like the tick of True Sight and Melf's Acid Arrow aren't doubled by haste, because it uses a genuine timer rather than a per-round basis. I believe haste cuts the combat round in half, which you can check by setting an auto-pause option for end of round (which triggers every 3 seconds in hasted combat, rather than 6, if using default framerate). Also inspite of the hardcoded 5 attacks per round cap a creature can have up to 6 apr under regular haste (provided they had 5 before casting). The balance issue of the Regeneration + Haste can certainly be solved with multiple delayed 17's. Improved Haste is the real game-breaker, and comes with it's own hardcoded stat for AI purposes - possibly it could be balanced with a shorter duration or even a mild damage penalty (you can use the Kai opcode with a negative value to avoid wraparound bugs) to reflect that striking twice as frequently necessarily means striking less accurately and forcefully.

SR already changes Blade Barrier and Insect Plague to work through multiple 146 sub-spells, so haste won't affect them, no? I suppose the same could be done for the Regeneration spell, using multiple delayed instances of opcode 17 in place of the regular opcode 98. In the case of "natural" regeneration and poison, I don't think a doubled rate is unbalancing or unrealistic because (1) the latter adds a "risk" to haste, normally an excellent spell (2) natural or equipped-item-regeneration is usually pretty slow (3) haste is mentioned to speed up your metabolism, presumably including rates of both healing and disease. Slow is actually a considerable debuff for characters with only 1 apr, because they're probably spellcasters and it doubles the casting time of any spell. Also, using any opcode besides 16 for haste won't correctly set STATE_HASTED for AI purposes, same problem I run into when trying to change the overpowered blindness effect.

Looking at the video, I believe Stworca is using "Fiends cast as normal wizards" option, which isn't recommended as many of their abilities such as Fire Storm, Symbol Stun and Blade Barrier have a long casting time and are easy to interrupt, considering most fiends lack defensive buffs. Btw. DavidW, I've found a more reliable way to simulate innate casting than SpellNoDec + casting speed bonus, as even spells with an alleged casting time of zero can sometimes be disrupted by damage (or silence, miscast magic effect etc.). Have the creature ForceSpell() an innate (casting time 1) which does nothing but display the casting graphic/animation, then in the same block ReallyForceSpell() the actual spells - the ReallyForceSpell() action is normally carried out under timestop or disabling effects, but will not be if the creature first has to force cast another spell in the same scripting block - although ForceSpell() does seem to be possible for unconcious creatures, STATE_SLEEPING would need to be checked for. Theoretically, Mel's stoneskin (once you're aware of this) is easier to remove than regular stoneskin, as it can be dispelled by Secret Word rather than Breach (expending a 4th level rather than 5th level slot), I'd support changing it to a combat protection.

You could increase the missile fraction of damage (to distinguish from fireball), but for a creature hit by multiple flame arrows in a round allow a second saving throw for the fire portion. This would make FA a bit stronger against groups of creatures, but not overpowered against a single creature. It's mostly useful for illusionists who cannot cast Spirit Armor, but I think it's fine as it is. The save penalty would need to be increased according to target's level, otherwise, this is a cheaper version of SI:Divination - a high level wizard with buffs could easily have a save vs spells of -6 or so. The 2nd ed version is actually mentioned as reducing missile damage by 1 pt per die (about 15%) even against magical missiles, which could be added. The vanilla spell is actually reasonably powerful if used in the right combination. I.e. a mage with ProMW + ProNM + teleport field is very difficult to attack. The problem is not that ProNM is underpowered but that elemental arrows are too effective at interrupting mages - currently only the fire arrow allows a save against the elemental damage. For the same reason I'd vote for "Neutralize Poison" to be reverted to curing blindness - I think various AI scripts still try to do this.

The standard "create magic item" opcode always uses magic weapon slot and can't equip a character for dualwielding, there is another code (143) which can, but is bugged and actually replaces whatever was in that slot permanently. Chill touch - make it work as a "melee hit effect" (i.e. monks' stunning blow) - so a character isn't "stuck" using CT if they run into something it cannot hit. Shocking Grasp, Ghoul Touch (the latter is supposed to be single target paralysis, but with a sickening "aura" that can get multiple enemies) etc. There is a way to make these usable only once but allow multiple attempts to hit a creature (that they would vanish on a miss is their biggest downside in vanilla) - rather than using expendable charges include a "remove item" 112 opcode under each touch attack's melee header so that it will remove itself on a successful hit.