Angband 4.0beta status

[Thraalbee]

Crash bug. New character in village. Rogue key set.
~
"g"-"m" displays are offset by one and "n" crashes the game.

[AnonymousHero]

Could you elucidate on what you mean by this, perhaps?

Hopefully. The idea would be to cache intermediate calculations in the p_ptr struct and to build a sort of hierarchy of recalculations triggered by patterns along the lines of "x_shadow != x, therefore recalc x and anything depending on x, afterwards set x_shadow = new x". Does that explanation make sense?

[MattB]

Too lazy to read backlog to see if this has been reported already, but you might want to look at this winning comp dump wearing the Massive Crown -- the stat display is kinda hilarious http://angband.oook.cz/ladder-show.php?id=17796. (It's entirely possible it looked like this before 4.0, also too lazy to find ladder dumps to corroberate.)

(congrats MattB)

What do you mean? My stats were like that from the start of the game. I just thought it was a 4.0 thing?


Seriously though, thanks debo and Ingwe for the congrats.

[MattB]

3. Never did get a confirmation that the change of the "Summons ringwraiths" in 4.0 was intended. In 3.5.1, when all the ringwraiths were dead, "summons ringwraiths" would instead actually summon fiends of darkness, but now "nothing comes". The change makes sense, but would just like confirmation that it was intended.

I think this is a good thing. But maybe we could have more flavour - "Morgoth howls with rage as no one answers."

[Timo Pietilä]

Summon uniques, however, if I recall correctly, if ALL of them were dead (well except Farmer Maggot and/or Father Christmas) then indeed nothing would come. A small reward for hunting every one of them down.

IIRC that used to be Jabberwocks then Black Reavers. At least in some point of angband development.

Many, many times more dangerous than most of the uniques.

[takkaria]

Maybe one solution would be to actually just do "caching" on the p_ptr struct (via shadow fields) instead of "hoping" that everything does the PU_XXX thing correctly? At least you'd have isolated the potential problems to a smaller bit of code.

Could you elucidate on what you mean by this, perhaps?

[AnonymousHero]

Sadly, I think that's not actually the case - there has been some noticeable slowdown in places, in some cases due to design decisions of mine based on the principle that computers are fast enough now...

Maybe one solution would be to actually just do "caching" on the p_ptr struct (via shadow fields) instead of "hoping" that everything does the PU_XXX thing correctly? At least you'd have isolated the potential problems to a smaller bit of code.

(I know it's easy to suggest these things when you don't have to implement them.)

EDIT: Actually, maybe just forget that: Try doing to the change and see if there's anyone who objects in practice. It's far easier to optimize correct code than it is to correct optimized code.

[Nick]

On this note: Are different bits of the code still using the "or a flag in p_ptr with PU_XXX mask" thing which restrict exactly which bits of the player are going to get updated? If so, I'd recommend just removing that for a general "update player" flag instead. I submit that all computers (even phones!) at this point are fast enough to not care about individual flags. Just recalculate and redisplay everything. If there's a problem with display speed, you just defer to the display code to avoid "flushing" things that have actually changed. It just soooooo much easier to avoid this kind of bug by doing this.

Sadly, I think that's not actually the case - there has been some noticeable slowdown in places, in some cases due to design decisions of mine based on the principle that computers are fast enough now...

[AnonymousHero]

Angband 4.0 dev 4c26520

That damnable Speed not updating until [^R]edraw is still prevalent. Damnit!

On this note: Are different bits of the code still using the "or a flag in p_ptr with PU_XXX mask" thing which restrict exactly which bits of the player are going to get updated? If so, I'd recommend just removing that for a general "update player" flag instead. I submit that all computers (even phones!) at this point are fast enough to not care about individual flags. Just recalculate and redisplay everything. If there's a problem with display speed, you just defer to the display code to avoid "flushing" things that have actually changed. It just soooooo much easier to avoid this kind of bug by doing this.

EDIT: I did this for player/dungeon display in T2 (PW_xxx, I think it was) and it made absolutely no observable difference in display update time. I'm planning on removing the PU_XXX thing, but the T2 code is... tangled.

[Ingwe Ingweron]

Angband 4.0 dev 4c26520

That damnable Speed not updating until [^R]edraw is still prevalent. Damnit!

[topazg]

Only one very minor bug I've found at the moment.

I had three stacks of pebbles (+0, +0) in my quiver, totalling 81 pebbles, taking up two quiver slots of 40 and one slot of a single pebble. I destroyed one pebble stack with "ignore all of these", and it cleared both of the two quiver slots of 40 and left the single pebble, now in a stack of 1, with the ignore flag but not removed.

Enjoying some of the changes, particularly the level feeling display at the bottom.

[Ingwe Ingweron]

....Failed summon of ringwraiths or uniques now gets greater undead as intended....

I'm pretty sure it used to be fiends of darkness in place of summon ringwraiths, wraiths and Vampires are what would show up if the ringwraiths were dead. Not the Greater Undead, which is a little more cruel. But that may be the intention.

Summon uniques, however, if I recall correctly, if ALL of them were dead (well except Farmer Maggot and/or Father Christmas) then indeed nothing would come. A small reward for hunting every one of them down.

[Ingwe Ingweron]

Too lazy to read backlog to see if this has been reported already, but you might want to look at this winning comp dump wearing the Massive Crown -- the stat display is kinda hilarious http://angband.oook.cz/ladder-show.php?id=17796. (It's entirely possible it looked like this before 4.0, also too lazy to find ladder dumps to corroberate.)

(congrats MattB)

Yeah, it always does that if @ don's the crown before the dump since the crown super-maxes all stats. Most people submit the dump before wielding the crown.

And definitely congratulations to MattB!

[debo]

Too lazy to read backlog to see if this has been reported already, but you might want to look at this winning comp dump wearing the Massive Crown -- the stat display is kinda hilarious http://angband.oook.cz/ladder-show.php?id=17796. (It's entirely possible it looked like this before 4.0, also too lazy to find ladder dumps to corroberate.)

(congrats MattB)

[Nick]

Update 4c26520 fixes:

• Knowledge screen commands for stores
• Improved trap descriptions
• Symbol lookup for 'P' now gets giant, not Morgoth
• Improve bless effect description
• Improve message for trying to walk into lava
• Make a (distinct) sound for picking up to or dropping from the quiver
• Return of the blank line under the object menu
• Failed summon of ringwraiths or uniques now gets greater undead as intended

I think that's all except:

• The display delay on missile shooting, which I think is just a thing now (it's a result of disconnect between the core function firing the missile and the display function displaying it)
• The chest mimic thing. That seems actually to be working as intended - the chest mimic is clear (like a clear hound), so it will look white, or yellow if you're using yellow torchlight.

Also of course there are the ten official bugs remaining.