Some thoughts about spells

[awldune]

I once wore a double breath from Glaurang when trying to cross a large room to get into a corridor to fight him. Took me about 2 months to get that pally competitive and I've been paranoid of the big breathers ever since... also haven't tried to fight Glaurang since. In my current game I think I can take him down now though.

Remember to always get a double resist (permanent + temporary) when facing heavy fire/cold breathers.

[awldune]

It only works adjacent to you. It is kind of radius 1 ball spell centered at you, not LoS like a staff or mage better of two spells.

Hard to imagine someone deciding that regular Sleep Monsters would be overpowered... I wonder if it used to work better in some old version of the game.

[bulian]

Just played a hobbit mage (comp char). A few thoughts:

MB8 looks rather silly for mages with only 2 spells, one of which is rather exploitive (brand). I didn't check but I think rogues and rangers have access to all of the spells in this particular book. Considering mages only get 4 blows per round, melee damage is pretty low to begin with - why were heroism/berserk removed from mages but left for the classes where the spells are not very useful?

At the end game, carrying books 1-7,9 was necessary as at least one spell was needed from each book:

MB1
Phase door

MB2
The occasional teleport self

MB3
Identify

MB4
Haste self

MB5
Resist

MB6
Rift

MB7
Recall/Tele level

Its been a long time since I've looked at paladins or priests, but late game priests can get away with far fewer books while maintaining the same utility due to redundancy within the spell set. I wouldn't be opposed to buffing mages slightly by adding a couple redundant spells to dungeon books.

bless durations are cumulative. So are berserk and hero. Unless that has been changed recently? Haste and resistance are not, for who knows what reason.

Resistance and shield spells are cumulative as of 3.3.1. Haste is not. I'm actually OK with haste not being cumulative as an exception since speed is so powerful, but recasting the spell should reset the duration and not add a piddly +5 turns or whatever the current behavior is.

Mana Storm (really needs to get to zero fail)

When I switched from rift to manastorm, things started dieing significantly faster. Mana storm crushes the alternatives. The main problem with manastorm and MB9 in general is its such a large jump in power. The same is true of MB6. Finding MB9 is almost a prerequisite for winning the game as a mage. No other class is as dependent on drops as mages.

IIRC at level 50 and max INT its failure rate is something like 9% or 13%,

Would it be unreasonable to remove the CL dependence of failure rate on all spells above a certain CL, such as CL35? Otherwise grinding a few levels to reduce failure rate is encouraged.

[artes]

All elemental ball spells, shards, meteors etc. destroy items at floor. There also is (too) wide variety of them, which causes annoyance in picking the right one (you can do only so much macros).

When I played a mage I actually liked to destroy loot, because I was too lazy to identify everything when looking for artifacts. (Artifacts were not destroyed. I don't know if that has changed.) Though the number of spells that destroy loot can be reduced, because I only need one, ha ha.

[kaypy]

Rend soul and chaos strike. I don't recall what these do, so I probably have not used them much. Weaker than other same manacost spells?

Rend soul is a nether bolt. Half damage vs evil makes it worthless against huge swathes of opponents, but against eg Angels it is your most effective spell for a reasonable time.

Chaos strike is competitive with manastorm, possibly right to the endgame, depending on the exact failrates. The problem with chaos strike is that it may polymorph the target, which also heals all damage. Thus you really only want to use it vs lone uniques.

Attached is what I believe is from the endgame of my last 3.3 mage playthrough. (I've been being brutally murdered in frogknows for a while, so my recollection is a little hazy). It is damage per mana vs damage per turn, including failrates and beaming.

[smileyy]

I'm assuming no one thinks that a caster is juggling 9 spell books and picking one out, opening it to a page and casting a spell from it.

Given that assumption, is the dependence on spell books for casting presuming something like reading them while resting, or other implicit activity? Or is it simply a means of picking the spell to cast?

I know Angband and realism (2H sword, shield, torch all at once) don't always see eye-to-eye, but I'd like to hear others thoughts about this.

[Derakon]

Kaypy: do you have a larger-resolution version of that image? I can't read the spells.

Smileyy: casters already have an easier time of inventory management than non-casters; removing the books entirely would make it too easy. I have always assumed that the casters are flipping to the appropriate page as needed, not that I've gone out of my way to justify that belief.

[kaypy]

Kaypy: do you have a larger-resolution version of that image? I can't read the spells.

I'm sure it was bigger before I uploaded it...

OK, heres a version where I've zoomed in to the spreadsheet a bit more before screendumping it.

...

AHA! It *does* shrink the attachment! Right- Here's a zip of the image. Shrink *that*, forum software!

spellgraph.zip

[Derakon]

Awesome, thanks.

You didn't take into account your device skill for the wands. I don't remember the exact math but a wand of annihilation should beat manastorm for average damage over time in the end.

It's interesting how many of the spells are clustered into this ball of "not really useful past the midgame". Also interesting that fire/ice and lightning/acid have the same efficiencies.

[kaypy]

You didn't take into account your device skill for the wands.

Hmm. I thought I had done that at some stage. Must of been one of the variants' spreadsheets...

It's interesting how many of the spells are clustered into this ball of "not really useful past the midgame".

Earlier, it tends to wind up with 'use whats on the perimeter'- trading off dps against manacost as appropriate, but by the end, the mana efficiency of MS isn't really that much worse than the element bolts...

Also interesting that fire/ice and lightning/acid have the same efficiencies.

That angle basically comes down to mana cost and failrate. Since this is endgame, the failrate is mostly 0 and its just a matter of those spells having the same cost.

As a comparison, here is l45 iq 18/150. Here you see the usecase for rend soul if you've still got any pesky non-evils to deal with. (Assuming Keleks has turned up)
spellgraph.zip

[Magnate]

Working on it. I have a branch on github where this works. However, it took considerable work to rewrite this, and I'm paranoid that I might have broken something in the process. The problem lay in that it was impossible to query the device damage without making the effect. The damage was hardcoded into the effects. I got around this by extracting the damage to list-effects.h with the eventual goal of making function pointers and removing the giant switch statement in effects.c altogether.

I actually need some dev guidance on whether my approach was a good one or not. But most everyone is busy with real life right now, so it'll have to wait.

Sorry, been on holiday ;-)

I had a long discussion with takkaria about effects.c, some of which is captured in http://trac.rephial.org/ticket/1427 and http://trac.rephial.org/ticket/1041. Re-reading it now, it looks remarkably like what I've just done with item affixes: it suggests a low-level set of single effects (cure some hp, set/clear a timed status, etc.) and then a higher-level grouping of compound effects (e.g. CCW cures hp, poison, confusion etc. etc.).

This means that each single effect would take one variable parameter, which would be the damage after adjustment for device skill. If we do this, you only have to calculate the damage to display it, you don't actually have to call the effect.

I haven't had a chance to look at your branch yet, but if it's consistent with that direction of travel, I'll happily merge it so that people can see adjusted device damage (or the multiplier).

[GeoffHill]

Re the elemental bolts.

I've always liked the progression feel as one builds levels in a mage, so I'd like to see them stay (maybe with damage tweaks if needed).

A wide ranging detection would be good for mages.

I think I like the fact the books contain spells that are less than optimised, but might have niche applications. Isnt that part of the appeal of a spell/prayer user? That moment when you are bleeding and have 5hp without potins and you look at every book you have and go... err... alter reality maybe?