Bug - Fixed Steely-Eyed Squint is being counted twice for +item, causing infinite bloat

[phreddrickk]

r26198; I've had at least two other users report this issue to me.
I have to attach some images; my mafia became more or less unusable:

After restarting mafia, another modtrace gave me:

Across all of these, I'm seeing the commonality of Squint appearing twice and giving insane quantities of +item; I suspect that it is doubling itself, and it is doing so repeatedly. I'm not sure what causes it to recurse. I'll scour my logs in a moment, but I figured I'd get this up sooner rather than later.

 

[phreddrickk]

Based on the fact that it was tamped back down when I logged back in, I think that it doesn't immediately jump into infinite descent, but instead doubles or quadruples or so every time that it gets re-evaluated.

 

[katyarn]

Perhaps the issue is it repeatedly gets copied here in PR #530? (Line 1666 to 1673 of Modifiers.java)
The Add function has a bit of a hack to ignore champagne bottle which Set lacks.

 

[fractalnavel]

Is the following related, maybe?:

 

[gausie]

Huh ok. I'll fix this.

 

[gausie]

It turns out mod() is a very weird function in ModifierExpressions

Mods doubling selves by gausie · Pull Request #551 · kolmafia/kolmafia

Fix bug where self-doubling effects were tending to infinity

github.com

 

[gausie]

Ok this is merged.

It actually turns out that mod() doesn't do anything at all, and all the effect doublers are defined at the end of recalculateAdjustments manually. I suppose I could refactor the effect-based ones to actually look at mod as intended there? But we should either make use of it or remove it, because this was really confusing.

 

[lostcalpolydude]

Ok this is merged.

It actually turns out that mod() doesn't do anything at all, and all the effect doublers are defined at the end of recalculateAdjustments manually. I suppose I could refactor the effect-based ones to actually look at mod as intended there? But we should either make use of it or remove it, because this was really confusing.

I agree that the code was confusing, when I made it work (I haven't looked at it in a while). To the point where I was surprised that having two doubling effects worked correctly, but since it was working I stopped touching it.

 

[gausie]

We just need to remove that modifier expression, it currently always returns zero and then we handle it manually at the end of recalculateAdjustments.

 

[lostcalpolydude]

I feel like some of the code was there to handle predictions (from the Maximizer) as opposed to current calculations for effects on hand.

I haven't set up an IDE since the transition to github, so I'm not in a position to even try looking at it.

 

[gausie]

Both calculated at the end of recalculateAdjustments, not with mod(), as far as I can tell.