Tinkered a bit more with this, seems the crash occurs whenever the kobold shaman casts a magic missile.

Made a new save just before the encounter. In the save file just walk north and walk around in the room until the kobold casts a magic missile. Have a bunch of the new and improved CLW, so it's quite safe :-)

That's odd, because I fixed a crash on magic missile earlier on in the spell refactor. Either I messed up and regressed somewhere, or it's a coincidence and it's a completely different problem.

Thanks for the savefile - I'll have a look this evening.

So far it's crashed for me each time someone cast a magic missile on me, so guessing it is still buggy...

I'm getting about the same: magic missile = destroy the world

I have seen it occasionally get the spell off without crashing, but it's 80% crashing so far...

Hmmm. Inconsistency makes bug hunting much harder, but if it happens 80% of the time it's probably going to be easy enough to find and fix.

Looks like there will be a new nightly out soon, as there's not much testing that can be done if magic missiles crash the game ...

Been doing some digging.

Should monster_learn_resists or update_smart_learn abort where
gf_ptr->resist
gf_ptr->immunity
or
gf_ptr->vuln
is null?

Because as near as I can tell, when that null goes into FLAG_OFFSET(flag) we get a negative-wraparound on the unsigned flag_offset which is what causes the crash.

Excellent, thank you - that explains the regression. Last time I had to add the nullness check to check_for_resist, this time it needs adding to update_smart_learn.

I'm not sure why the assertion failure was in z-bitflag.c though, when there was already an assert in update_smart_learn itself. Compiler optimisation, I guess.

Anyway, a new nightly should be out in a couple of hours. Thanks for the catch.

Yet another build
 

Fixes for the tunneling bug that kaypy reported and for the magic missile crash that jens reported are now on their way to the next dev release. Note that the one marked ef9e70e6dc still has some problems; you'll want the one marked 31b20dac9b. It should be up in a couple hours. The only other changes from the 22 May 2011 release are the Mac OS X accented characters fix, some tweaks to level feelings, and a more sensible message at death about keeping randarts.

The assert used to be assert(what >= 0), which wouldn't catch the null...

Totally agree With Timo.

Cambeleg and FIngolfin are top artifacts--I want to see these restored.
Same with the rings of power--I totally welcomed their recent boosting in speed and damage because in my mind they always had trouble competing with rings of speed, when they really ought to be something special--tip top artifacts.

I sort of disagree stripping speed from Gondor--in my current game I totally don't find Gondor a no-brainer over Dor-Lomin because of the letter's telepathy. Gondor is a top artifact.

If you want to tone down Elessar, Numenor, Haradrim, and Hammerhand that is totally reasonable--Haradrim is a new nontraditional artifact anyway, and I really don't like all these artifacts of evil no-class Harad being so powerful relative to the far more glorious Elven, Dwaven, Westernesse, and Godly origin stuff--the Haradrim just weren't portrayed as that powerful. Elessar--well, it's kind a newish anyway and I think a bit too powerful, but how about toning doen a *little*--stripping speed and removing all 7 damage is too much. Hammerhand--well, +9/+9 is too powerful especially as Helm Hammerhand just wasn't godlike, though he is certainly remembered fondly by the Rohirrim. So take it down a bit *but* also get rid of *aggravates*--it will at least be useful then.

I won't be downloading this development version.