V 3.5 now in feature freeze

[takkaria]

Also, for the second time in a week I've had a crash when inscribing some arrows with {=g @f9}. Unfortunately I wasn't able to reproduce it last time so I didn't report it. But as it crashes so rarely, twice in a week can't be coincedence, can it?

(Mind you, last week I managed to summon 7 up stairs in a row while looking for a down stair. My personal record and odds of about 100,000 to one!)

Hm, thanks. Is this with a 'display inven/equip' term window?

[MattB]

In the day? It's because the sun is up.

Well then, that's OK then!

[MattB]

Hm, thanks. Is this with a 'display inven/equip' term window?

yes it is. But the action was done by pressing <{> first (rather than <e>, <n>, <{>).

[With apologies for the punctuation train crash!].

[MattB]

2^7 is only 128, so it's not that unlikely.

Sorry, maths crash!
I powered up the fail rate instead (as that's normally the relevant stat in a long chain of undesirable outcomes). I'm so ashamed.

As for torches, the game is just not penalizing you for taking your light source off when there's no need for it. You'll still use up turns if it's night time.

I guess I never noticed that effect! It's funny what you notice when you're actuvely looking for bugs.

[Nomad]

Okay, here's another tiny fix for an error in the room_template.txt file. (Maze layout that you can't actually get all the way around. Oops.)

[Timo Pietilä]

I guess I never noticed that effect! It's funny what you notice when you're actuvely looking for bugs.

It's new. I didn't know that either, and old versions definitely used light even when you didn't need it if you had lightsource equipped. I have no idea when that was changed.

[PowerWyrm]

Also, for the second time in a week I've had a crash when inscribing some arrows with {=g @f9}. Unfortunately I wasn't able to reproduce it last time so I didn't report it. But as it crashes so rarely, twice in a week can't be coincedence, can it?

(Mind you, last week I managed to summon 7 up stairs in a row while looking for a down stair. My personal record and odds of about 100,000 to one!)

The crash has already been reported. Myself, I've never been able to reproduce that.

Concerning stair creation, the RNG is known to be seriously biased (same problem with teleport: when you use teleport twice, you almost always end up very close to where you were). When you create stairs deep in the dungeon, it's really common to generate 7-8 consecutive up stairs. That's why for my variant I added a little bit of code to ensure you'll never get more than 5 consecutive up or down staircases (and also ensure that double teleport will never bring you back on the same spot).

[PowerWyrm]

It's new. I didn't know that either, and old versions definitely used light even when you didn't need it if you had lightsource equipped. I have no idea when that was changed.

Err... I just checked and this has been changed 6 years ago (Jun 06, 2007)! This sounds like this is the normal behavior since 3.0.8...

[Timo Pietilä]

Err... I just checked and this has been changed 6 years ago (Jun 06, 2007)! This sounds like this is the normal behavior since 3.0.8...

New for me. Anything starting with 3 is new

[takkaria]

Concerning stair creation, the RNG is known to be seriously biased (same problem with teleport: when you use teleport twice, you almost always end up very close to where you were). When you create stairs deep in the dungeon, it's really common to generate 7-8 consecutive up stairs. That's why for my variant I added a little bit of code to ensure you'll never get more than 5 consecutive up or down staircases (and also ensure that double teleport will never bring you back on the same spot).

The new RNG really shouldn't be biased, because it's meant to be a very good PRNG... I wonder if there's any way to reduce this effect without adding explicit checks against it.

[MattB]

The new RNG really shouldn't be biased, because it's meant to be a very good PRNG... I wonder if there's any way to reduce this effect without adding explicit checks against it.

I don't really know what I'm talking about here, but couldn't it be set up so that in genuine 50/50 situations it has a 60% chance to offer the opposite outcome after the first event?

Although I'm struggling to think of any other genuine 50/50 chances...
Teleport level, I suppose, but you're unlikely to cast it successive times.

[Derakon]

The problem with Teleport Other is more down to how the spell is coded than how random the RNG is. The spell aims for plunking you down within a certain distance range from your current position, and unfortunately given Angband's dungeon topography, that tends to mean there's only a small handful of possible locations where you can land up. And if you teleport from your new location, your old location is a valid target.

My plan for Pyrel is for Teleport Other to be able to land you any distance up to the maximum distance, rather than for it to try to force you to be at least some minimum distance away. Which means that sometimes you'd cast the spell and it would have an effect like Phase Door, but I suspect it will also be much more reliable as an escape. It could well be worth trying in Angband proper.

[Timo Pietilä]

The problem with Teleport Other is more down to how the spell is coded than how random the RNG is. The spell aims for plunking you down within a certain distance range from your current position, and unfortunately given Angband's dungeon topography, that tends to mean there's only a small handful of possible locations where you can land up. And if you teleport from your new location, your old location is a valid target.

(Teleport self, not other). It used to be much worse. Teleport had so long range at the high levels that you basically found acceptable positions only at the furthest away corner of the dungeon. Priest portal was much better. Now teleport has lower minimum range for acceptable positions so it isn't quite that bad but it still tends to jump you between two corners.

[Timo Pietilä]

The new RNG really shouldn't be biased, because it's meant to be a very good PRNG... I wonder if there's any way to reduce this effect without adding explicit checks against it.

RNG isn't biased, human mind is. You remember extraordinary things much better than ordinary things, and bad extraordinary even better than good extraordinary.

For stair creation you never see series of positive outcomes because you take the first stairs that match your preference. So you only see those kind of series only for negative outcome. You would need to remember all the cases where you got the positive outcome at first try in order to find out that there are similar series of luck in other direction as well. Unfortunately getting positive outcome at first try isn't extraordinary so you forget.

[Ingwe Ingweron]

Is it a bug that if you choose location targeting, but a monster happens to be on the location, the targeting reverts back to monster targeting? It seems to me that when I choose to target a location rather than a monster, that I really mean to target the location. As it currently stands, if a monster happens to be there when I target the location, the targeting reverts to targeting the monster and when I kill it, I have to retarget the space. Anyone else find this frustrating?