00:00:32 <Eksell> Unstable branch on CRAWL.XTAHUA.COM updated to: 0.24-a0-108-g011bc4b (34)
00:05:17 <alexjurkiewicz> !lg * t won
00:05:19 <Sequell> 686. elliptic the Ruinous (L27 FeWz of Sif Muna), escaped with the Orb and 4 runes on 2019-02-14 04:49:01, with 1648328 points after 143721 turns and 6:13:29.
00:05:20 <alexjurkiewicz> !lg * t zot
00:05:21 <Sequell> 145. Aeraethus the Harbinger of Death (L23 DsGl of Yredelemnul), blasted by a red draconian annihilator (crystal spear) on Zot:3 (grunt_deathspiral_small) on 2019-02-14 04:32:12, with 504432 points after 65850 turns and 5:15:01.
00:05:29 <alexjurkiewicz> !lg * t0.22 won
00:05:30 <alexjurkiewicz> !lg * t0.22 zot
00:05:51 <Sequell> 2148. Yermak the Warrior (L21 DDFi of Makhleb), escaped with the Orb and 3 runes on 2018-08-26 19:57:36, with 6131627 points after 10243 turns and 2:07:13.
00:05:51 <Sequell> 343. WaifuGate the Conqueror (L27 GrBe of Trog), mangled by an Orb Guardian on Zot:5 (hall_of_Zot) on 2018-08-26 19:29:07, with 613556 points after 80110 turns and 3:19:29.
00:06:04 <alexjurkiewicz> 0.21 vs 0.16
00:06:28 <alexjurkiewicz> pretty naive comparison but pretty big difference
00:08:06 <gammafunk> not actually sure what you're comparing
00:08:25 <gammafunk> !lm !bot !boring t0.22 br.enter=zot / lg:br=zot
00:08:30 <gammafunk> looks like what you want
00:08:39 <gammafunk> !lm !bot !boring t br.enter=zot / lg:br=zot
00:09:55 <Sequell> 90s limit exceeded: killed !lm !bot !boring t0.22 br.enter=zot / lg:br=zot
00:10:22 <Sequell> 145/896 milestones for bot (!boring t br.enter=zot): N=145/896 (16.18%)
00:11:10 <gammafunk> !lm !bot !boring t0.22 br.enter=zot / lg:br=zot
00:12:04 <gammafunk> the difference is significant but definitely not drastic either
00:12:27 <Sequell> 90s limit exceeded: killed !lm !bot !boring t0.22 br.enter=zot / lg:br=zot
00:12:31 <gammafunk> !lm !bot !boring t0.22 br.enter=zot / lg:br=zot
00:12:37 <gammafunk> come on sequell, you can do it
00:14:04 <Sequell> 90s limit exceeded: killed !lm !bot !boring t0.22 br.enter=zot / lg:br=zot
00:14:16 <gammafunk> weird how it's struggling with relatively few games
00:14:18 <gammafunk> !lm !bot !boring t0.22 br.enter=zot / lg:br=zot
00:14:41 <gammafunk> but I checked this earlier in tourney, and it was like 12% or 14%?
00:15:49 <Sequell> 90s limit exceeded: killed !lm !bot !boring t0.22 br.enter=zot / lg:br=zot
00:26:28 <alexjurkiewicz> !lm !bot !boring t0.22 br.enter=zot
00:26:30 <Sequell> 2727. [2018-08-26 19:52:01] igooo the Executioner (L22 DDFi of Makhleb) entered the Realm of Zot on turn 26748. (Depths:5)
00:26:44 <alexjurkiewicz> !lg * !bot !boring t0.22 zot
00:26:45 <Sequell> No keyword 'bot'
00:27:06 <alexjurkiewicz> !lg !bot !boring t0.22 zot
00:27:06 <Sequell> 340. WaifuGate the Conqueror (L27 GrBe of Trog), mangled by an Orb Guardian on Zot:5 (hall_of_Zot) on 2018-08-26 19:29:07, with 613556 points after 80110 turns and 3:19:29.
00:27:37 <alexjurkiewicz> 12%.... right?
00:27:45 <alexjurkiewicz> stats is hard
00:28:24 <alexjurkiewicz> 33% difference. Same as my test :P
00:54:38 <gammafunk> yeah, it's just that the milestone query reports it directly...that is if it actually executes
02:02:28 <Cheibriados> Sacrificing a hand made my {*Contam} ring jump to the other finger. 13https://crawl.develz.org/mantis/view.php?id=11879 by Yermak
02:25:56 -!- amalloy_ is now known as amalloy
03:52:01 <Kelbell> Fork (bcrawl) on crawl.kelbi.org updated to: 0.23-a0-1718-g3e42c2ff6c
03:57:36 -!- amalloy is now known as amalloy_
04:53:25 <dontgetcocky> hey guys how do you replay a specific timeframe in a specific replay?
04:53:38 <dontgetcocky> using the irc commands?
04:53:52 <dontgetcocky> I just witnessed the greatest survival of all time
05:00:34 <Lantell> dsol (L2 GrEE) ASSERT(level >= 0) in 'mutation.cc' at line 462 failed. (D:1)
05:03:23 <dontgetcocky> nvm
09:11:07 <advil> !crashlog
09:11:18 <Sequell> 20690. dsol, XL2 GrEE, T:482 (milestone): https://underhound.eu/crawl/morgue/dsol/crash-dsol-20190214-100033.txt
09:12:44 <advil> "trample resistance: 0 (1 innate)"
09:13:52 <advil> haha, "Missing version history before: 0.13-a0-1804-g486adc7"
09:13:56 <advil> you don't say
09:17:45 <advil> !lg dsol gree
09:17:46 <Sequell> 1. dsol the Digger (L2 GrEE), slain by a hobgoblin (a +0 club) on D:1 on 2019-02-14 10:03:13, with 16 points after 507 turns and 0:05:24.
09:19:39 <Cheibriados> 03advil02 07* 0.24-a0-109-g9a59f91: Improve a very old fixup 10(47 seconds ago, 1 file, 5+ 1-) 13https://github.com/crawl/crawl/commit/9a59f91d9abe
09:32:46 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.24-a0-109-g9a59f91d9a (34)
11:57:30 <Cheibriados> 03advil02 07* 0.24-a0-110-gf57072b: Fix a couple of des cache random tag bugs 10(8 minutes ago, 2 files, 7+ 7-) 13https://github.com/crawl/crawl/commit/f57072b2d696
12:02:22 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.24-a0-110-gf57072b2d6 (34)
12:06:53 <Henzell> Unstable branch on crawl.akrasiac.org updated to: 0.24-a0-110-gf57072b (34)
13:36:09 -!- amalloy_ is now known as amalloy
13:45:09 <Henzell> reaver (L17 HOFi) ERROR: range check error (27000 / 2000) (Snake:4)
13:56:00 <Henzell> reaver (L17 HOFi) ERROR: range check error (27000 / 2000) (Snake:4)
13:58:51 <advil> !crashlog
13:59:07 <Sequell> 20692. reaver, XL17 HOFi, T:45076 (milestone): http://crawl.akrasiac.org/rawdata/reaver/crash-reaver-20190214-185559.txt
13:59:41 <Cheibriados> Game crashes when I try to bring up the "^" screen. 13https://crawl.develz.org/mantis/view.php?id=11880 by Reaver
14:04:24 <advil> a preemptive save backup, it's like this report is from a dev or something
14:04:46 <advil> what seems to be happening is that an orc has been polymorphed into a fire crab, and no longer has its items
14:06:23 <advil> or rather, a shapeshift
14:06:23 <advil> er
14:06:42 <advil> probably from a chaos effect
14:09:28 <gammafunk> advil: yeah, that report was was from reaverb, a dev
14:09:33 <advil> I know
14:09:40 <gammafunk> oh ok
14:19:41 <Cheibriados> 03advil02 07* 0.24-a0-111-gfd6172a: Fix Beogh crash for polymorphed followers (11880 / reaverb) 10(3 minutes ago, 1 file, 11+ 4-) 13https://github.com/crawl/crawl/commit/fd6172a7371f
14:25:28 <gammafunk> getting flashbacks to that old beogh follower polymorph bug
14:25:50 <gammafunk> where my save got corrupted so that I couldn't exit vaults since I couldn't go up to vaults:3
14:26:00 <advil> heh yeah it's the same setup conditions
14:32:45 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.24-a0-111-gfd6172a737 (34)
14:46:49 <Cheibriados> 03advil02 07* 0.24-a0-112-g3372e67: Uncomment an accidentally commented vault call 10(33 seconds ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/3372e67c9135
15:02:32 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.24-a0-112-g3372e67c91 (34)
15:10:14 <advil> kind of interesting and not entirely irrelevant to dcss: https://motherboard.vice.com/en_us/article/43zak3/the-internet-was-built-on-the-free-labor-of-open-source-developers-is-that-sustainable
16:37:00 <Yermak> FR: make Ignasio guaranteed by Gloorx floor, like Ilsuiw by Shoals:4.
16:40:39 <amalloy> Yermak: why? i'm not trying to speak for everyone here but i always found it puzzling that ilsiuw is like the only unique guaranteed to appear in the same place every game (though i guess hell lords fit this description too)
16:41:52 <ebering> Ilsuiw is no longer guaranteed on shoals 4
16:46:39 <gammafunk> yeah, ilsuiw's placement got changed in 0.22
16:47:22 <gammafunk> I don't think guaranteeing ignacio on a certain floor is a good idea, but that unique is also not a very well designed one, which is a bigger problem
16:49:17 <gammafunk> ignacio could certainly just be made more common
16:50:43 <alexjurkiewicz> !lg * recent killer=ignacio s=place
16:50:44 <Sequell> 6 games for * (recent killer=ignacio): 6x Pan
16:51:06 <amalloy> !lm * recent uniq=ignacio
16:51:18 <Sequell> 684. [2019-02-14 00:01:38] Bellerephron the Minotaur Barricade (L27 MiFi of The Shining One) killed Ignacio on turn 120817. (Pan)
16:55:32 <gammafunk> =nemelex FoSk
16:55:35 <Sequell> Defined keyword: nchoice => GhVM|TeEn|VpCj|VpAE|CeAE|DDIE|DEMo|FeWr|HaEE|GnHu|KoAK|FoSk
17:05:14 <alexjurkiewicz> advil: thanks for the article link
17:15:48 <Cheibriados> 03Umer Shaikh02 07https://github.com/crawl/crawl/pull/982 * 0.24-a0-113-ga5ec544: Make god gifts interrupt runrest delays only (11825) 10(7 days ago, 1 file, 3+ 2-) 13https://github.com/crawl/crawl/commit/a5ec5441c6d0
17:21:07 <Cheibriados> 03Umer Shaikh02 07https://github.com/crawl/crawl/pull/982 * 0.24-a0-113-g355c810: Make god gifts interrupt runrest delays only (11825) 10(7 days ago, 1 file, 3+ 2-) 13https://github.com/crawl/crawl/commit/355c8109e3fe
17:22:11 <alexjurkiewicz> lol https://crawl.develz.org/tavern/viewtopic.php?f=17&t=26085
17:31:16 <gammafunk> haha, that's amazing
18:05:06 <rax_> CAO is low on disk space again. I've identified some chaff I can get rid of this weekend that should do the trick for another 6-12 months with no downtime. I may also schedule a time to increase partition size that would require a few hours of downtime, but I'll give more warning before that.
18:05:38 <gammafunk> rax_: thanks for letting us know
18:08:34 <amalloy> i'm surprised muting the message helped
18:09:00 <advil> I'm not
18:09:00 <advil> I've definitely seen crashes when the message code gets overwhelmed
18:09:09 <advil> in fact this might be a way to reproduce that, which has always been a challenge
18:13:30 <gammafunk> I do wonder if dispersal traps going dormant for a few turns might work
18:14:18 <gammafunk> not sure, it's more complexity for traps
18:14:32 <gammafunk> could extend the idea to zot traps as well
18:15:08 <ebering> zot traps need a more general overhaul, but if that makes them deadlier then they'll probably want that dormancy
18:15:41 <advil> if I'm understanding the picture that's all happening on the same turn, right? not immediately retriggering might be a good idea
18:16:35 <ebering> yeah
18:16:43 <ebering> the recursive triggering is the problem
18:17:38 <gammafunk> ah, that would be ok
18:17:52 <gammafunk> I imagine it's a problem for zot traps too
18:17:58 <gammafunk> just vastly less common
18:18:26 <gammafunk> might be good if all traps that could move the player just avoided that recursion then
18:23:36 <Lantell> Unstable branch on underhound.eu updated to: 0.24-a0-112-g3372e67c91 (34)
18:26:14 <gammafunk> just randomly occurred to me that for all the weird morg ideas, it would probably work better if we just changed the weapon type completely
18:26:41 <gammafunk> especially since we have spriggan's knife, not really a need for a second unrand dagger that's not about the one exceptional thing daggers do (stabbing)
18:27:06 <alexjurkiewicz> https://github.blog/2019-02-14-introducing-draft-pull-requests/ <-- expect about a billion of these from me
18:27:24 <gammafunk> it could be an unrand staff of death or something
18:28:37 <gammafunk> just remember, alex: quality not quantity!
18:29:04 <alexjurkiewicz> bigger is better
18:29:18 <gammafunk> wow, you must have a 100% Og winrate
18:29:27 <gammafunk> !gamesby chequers og
18:29:28 <Sequell> chequers (og) has played 151 games, between 2014-09-29 11:50:18 and 2019-02-09 00:30:41, won 1 (0.7%), high score 1961156, total score 6685725, total turns 1385983, play-time/day 0:01:43, total time 1d+21:50:51.
18:30:14 <gammafunk> this is where I would use an emoji like 🐸
18:30:22 <advil> I wonder if we'll still get every email for draft PRs
18:30:34 <gammafunk> oh another thing that's new
18:30:35 <advil> though i can see the use case for it
18:30:42 <gammafunk> you can make commits out of PR reviews
18:30:54 <gammafunk> the commits it makes are pretty weird though
18:31:24 <gammafunk> you hit this button to make a commit in the PR with the suggested change
18:32:11 <gammafunk> https://github.com/crawl/crawl/pull/961
18:32:13 <gammafunk> has some examples
18:32:37 <gammafunk> helpful commit message of Update crawl-ref/source/dat/des/branches/swamp.des
18:32:46 <gammafunk> maybe that's the default and I didn't see where to edit it
20:27:11 <gammafunk> =nemelex DDFE
20:27:11 <Sequell> Defined keyword: nchoice => GhVM|TeEn|VpCj|VpAE|CeAE|DDIE|DEMo|FeWr|HaEE|GnHu|KoAK|FoSk|DDFE
21:20:21 <alexjurkiewicz> advil: are you interested in reports of seeded games that diverge?
21:20:30 <advil> sure
21:20:36 <advil> I may have fixed it if it's in 0.23 though
21:21:00 <alexjurkiewicz> hm. let me check I am using latest master. Might be a few days old
21:21:08 <advil> well, seed #s have changed
21:21:20 <advil> so if it doesn't reproduce that doesn't mean the bug is actually fixed, unfortunately
21:21:28 <alexjurkiewicz> :)
21:22:09 <alexjurkiewicz> ok, my build is a bit old. I'll try to find some new seeds that break
21:22:22 <advil> do you have a 0.23 seed number?
21:22:33 <alexjurkiewicz> no, I was just testing with an older trunk version
21:22:38 <alexjurkiewicz> %git e52b27f005
21:22:38 <Cheibriados> 07advil02 * 0.24-a0-99-ge52b27f: Place AK start in abyss with pregeneration on (11871) 10(3 days ago, 2 files, 13+ 9-) 13https://github.com/crawl/crawl/commit/e52b27f00579
21:23:12 <alexjurkiewicz> let me try to repro with latest trunk
21:23:21 <advil> ok
21:23:25 <advil> in what way did it diverge?
21:28:03 <advil> if it diverges on that commit it probably isn't fixed
21:28:25 <advil> but the unique placement change after that likely changed a lot of seeds
21:33:52 <alexjurkiewicz> d:10 had different vaults and all subsequent levels were changed
21:34:10 <alexjurkiewicz> this is with seed 5 and pregen off (I'm testing qw and re-running the first 100 seeds repeatedly)
21:35:35 <advil> so between instances of qw?
21:35:47 <advil> or were you using the lua test?
21:36:17 <advil> I should really run it with a lot of seeds overnight
21:39:36 -!- amalloy is now known as amalloy_
21:42:20 <Cheibriados> 03advil02 07* 0.24-a0-113-g597dbe2: Fix a math error in a test 10(68 seconds ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/597dbe2dff2d
21:47:53 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.24-a0-113-g597dbe2dff (34)
21:49:29 <alexjurkiewicz> advil: I was testing a change in qw which modified behaviour in zot. So unless something really funky happened, it seemed to be with "the same" qw
21:50:21 <advil> so the vault names were coming from qw games?
21:51:08 <advil> also, did you happen to record what the D:9 vaults were?
21:51:29 <advil> I can't replicate it with just the vault catalog so it might need a real game to trigger
21:51:52 <advil> btw if you find something like this and I'm not around I also opened an ongoing github issue to track things like this
21:54:53 -!- amalloy_ is now known as amalloy
22:02:35 <alexjurkiewicz> gotcha. I'm re-running qw now. The first few I'll probably report directly to you in case there are silly mistakes I'm making, to keep from messing up the issue
22:51:46 <travis-ci> The build was broken. (master - 597dbe2 #11138 : advil): https://travis-ci.org/crawl/crawl/builds/493567393
23:14:50 <Cheibriados> 03advil02 07* 0.24-a0-114-g5d539f0: shorten test again 10(86 seconds ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/5d539f04da09
23:17:26 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.24-a0-114-g5d539f04da (34)
23:18:39 <advil> alexjurkiewicz I'm getting surprisingly deterministic behavior for qw on that seed (in terms of what it does), but one thing I notice in both runs so far is that it went into lair and did most of lair after D:9. So  if something happened to cause it to do levels in a slightly different order, D:10 might come out different
23:19:06 <advil> e.g. if it went to D:10 first for some reason
23:19:19 <advil> (I wouldn't expect qw to be reliably deterministic like this all the time)
23:20:50 <advil> I guess if your version is modded it might do something different than mine, which is the standard one + ebering's patches
23:21:24 <advil> but anyways, probably the first thing to check, which won't be so readily apparent in the vault listing in a morgue, is the traversal order between runs
23:35:49 <alexjurkiewicz> yeah, will do
23:35:59 <alexjurkiewicz> but isn't each branch meant to have its own seed?
23:36:18 <alexjurkiewicz> so it doesn't matter if you generate d:8, lair:1, d:9 vs d8 d9 l1
23:36:18 <advil> yes, but vault generation is still very much not independent of what's already happened
23:36:20 <alexjurkiewicz> or is that where you're saying the bug is
23:36:24 <alexjurkiewicz> gotcha
23:36:30 <alexjurkiewicz> i'll get back to you
23:36:30 <advil> no, it's not a bug, this is why pregeneration is necessary
23:37:02 <alexjurkiewicz> oh, ok. You don't think crawl will ever have consistent incremental generation?
23:37:08 <alexjurkiewicz> (bbl)
23:37:12 <advil> for example, suppose that both D:10 and Lair:1 would each first try to place the same unique. Then, the first one that you enter will, and the second one will have a completely different set of maps, causing everything to diverge from there on
23:38:53 <advil> a lot of stuff would have to be restructured completely to really do that. I should eventually be able to implement more incremental pregeneration, doing it on demand if a player skips around
23:40:04 <advil> I tried to do that already and it was technically nightmarish without a rewrite of the level entering code, the first version had weird and inexplicable connectivity bugs
23:41:44 <advil> it would be possible in principle to have all random decisions specific to each branch. But for example that would rule out (or force one to decide ahead of time, independently of levelgen) letting uniques place in multiple branches
23:41:48 <advil> some for unique vaults and items
23:42:19 <advil> there might be tricks I can do to better partition some aspects of levelgen from each other, make it less sensitive to really minor things
23:54:17 <gammafunk> advil: isn't it impossible to let decisions affecting levelgen happen  ahead of time, since unlike with unique generation, you can't know if a level can even be made given the choices
23:54:46 <gammafunk> that is, if I try to choose a set of vaults ahead of to make D:3, I can't know if the D:3 I generate will veto
23:55:13 <advil> yeah, vetoes are a general problem with our system + better independence
23:55:35 <gammafunk> Although I guess you could just try to generate the level given those choices until it succeeded
23:56:12 <advil> one extreme solution (that would be better designed for from the ground up) is making all vaults specific to a branch
23:57:07 <advil> or other level range, doesn't have to be a branch, that's just the way the rng is currently partitioned
23:57:46 <gammafunk> doesn't that not fully solve the problem, since I could elect to shaft (or not) and get different visitation order within some branches?
23:58:16 <advil> yeah, but generating a few levels at a time is relatively fast
23:58:31 <gammafunk> oh
23:59:47 <advil> I  think some kind of generation on demand when you skip around from a canonical order might be good enough