00:17:17 Unstable branch on CRAWL.XTAHUA.COM updated to: 0.23-a0-180-ga5275e7 (34) 00:30:11 FR: smoke bomb spell so I can underuse it like I underuse scroll of fog 00:59:26 gammafunk: thanks, I'll check it out 01:50:15 -!- Mindiell_ is now known as Mindiell 03:14:41 Stable (0.21) branch on crawl.beRotato.org updated to: 0.21.2-0-g371e54c 03:26:29 Unstable branch on crawl.beRotato.org updated to: 0.23-a0-180-ga5275e7 (34) 05:02:52 http://crawl.develz.org/wordpress/0-22-1-and-0-21-2-bug-fix-releases 05:10:29 -!- amalloy is now known as amalloy_ 05:18:05 WebTiles: Using Shift-X [ can be used to reveal location within a new level 13https://crawl.develz.org/mantis/view.php?id=11681 by RoGGa 06:01:01 Walljump makes airborne attacks, but water makes them fail. 07:35:21 !lg * DE s=char / win 07:35:25 2997/690607 games for * (DE): 960/154794x DEFE [0.62%], 599/131157x DECj [0.46%], 443/120036x DEWz [0.37%], 139/22568x DEIE [0.62%], 120/17726x DEEE [0.68%], 71/9046x DEVM [0.78%], 57/15979x DEAE [0.36%], 56/22236x DESu [0.25%], 54/3767x DEBe [1.43%], 43/24165x DEEn [0.18%], 43/24878x DENe [0.17%], 39/6900x DEAM [0.57%], 35/3426x DECK [1.02%], 33/66487x DEWn [0.05%], 32/3009x DEMo [1.06%], 31/2803... 07:39:53 !lg* Gr s=char / win 07:40:08 !lg * Gr s=char / win 07:40:11 6294/438713 games for * (Gr): 2432/175418x GrFi [1.39%], 892/46075x GrBe [1.94%], 782/75548x GrEE [1.04%], 590/24671x GrGl [2.39%], 302/25647x GrMo [1.18%], 179/19721x GrFE [0.91%], 144/14204x GrWz [1.01%], 125/7794x GrWn [1.60%], 89/4657x GrCK [1.91%], 75/4960x GrCj [1.51%], 68/1565x GrAr [4.35%], 63/4359x GrSk [1.45%], 57/1609x GrAs [3.54%], 57/5688x GrVM [1.00%], 54/2503x GrHu [2.16%], 48/4232x... 08:01:33 advil: future bones file releases should definitely include all developer team names as easter eggs 08:01:33 if there is some way to make particular names extremely rare 08:01:49 maybe that should just be a hardcoded list in a *.cc file 08:24:34 maybe that should just be a hardcoded list in a *.cc file 08:24:35 oops 09:53:31 Hey devs! I'm working on a machine learning project as an undergrad "capstone project" using DCSS and qw as my playgorund. Optimizing the game for winrate seems difficult if the game takes as long as it does to play (around 10-20 minutes for the bot), so I was wondering if anyone knew if there were any artifical constraints of input delays in the games source code that could be removed to make the game play faster. I.E., only 09:53:45 so many ms have elapsed or something like that 10:00:24 duckroller: qw's rc (if not set up for online play) already turns off all delays, but 10-20 minutes seems long 10:00:25 i believe that elliptic's offline qw wins only take a few minutes, 10-20 minutes sounds long so i'm not sure what's up there 10:00:36 duckroller: there a few settings (set rest_delay, explore_delay, travel_delay to -1) 10:00:46 also, if you run it headless it goes faster 10:00:59 duckroller: Can we see the rc you're using for qw? 10:01:07 qw has some throttle settings internally that I'm sure you've found 10:01:22 * Pinkbeast has a 2h55m qw win but that was on a sloooow computer 10:02:12 also I'm not sure how up to date qw is for 0.22/0.23, so there may be conditions under which it gets stuck 10:10:14 thanks for the quick responses. I have set all the obvious delays in qw to -1. I'm running it headless on relatively modern architecture... And yeah, there are definitely some conditions that make it stuck, but those games will just go on forever rahter than take a while. I've worked out a few of them (BURN_SPELLBOOKS = False, make acquirement scrolls work in .22, etc...) but after running a bunch of trials overnight I'll sti 10:15:30 have you been using 0.22? 10:15:45 yes 10:15:57 I wonder if there's some new performance bottlenecks that no one has noticed yet 10:15:59 duckroller: Can we see the rc file? 10:16:06 sure, gimme a bit 10:16:16 ... and yeah, check if it's equally slow in 0.21 10:18:31 https://pastebin.com/4ndD2CWk 10:18:42 I'll look into running it on .21 in the near future 12:56:29 !tell gammafunk I think the windows console download might be misnamed relative to the downloads page (there's a file named just stone_soup-0.22.1-win32.zip in http://crawl.develz.org/release/0.22/) 12:56:30 advil: OK, I'll let gammafunk know. 13:12:48 -!- amalloy_ is now known as amalloy 13:18:07 Unstable branch on crawl.akrasiac.org updated to: 0.23-a0-180-ga5275e7 (34) 13:29:38 !tell gammafunk fyi just added the following line to the release post: "If you have previously installed 0.22.0, we recommend that you remove any files named `*.store.*` from the bones subdirectory of your saves folder, in order to get the full effect of this change." 13:29:39 advil: OK, I'll let gammafunk know. 13:31:01 only realized that issue just now; it won't use the dist_bones folder if there's a regular permastore with even one ghost, which locks the player into only ever seeing themselves for that level 13:31:38 that might be something I should change 14:42:06 !versions 14:48:35 &versions 14:48:41 one of these things 14:48:46 ??& 14:48:47 &[1/1]: Either a {pan lord} or a {hell lord}. 14:48:56 CAO: 0.23-a0-180-ga5275e7, CBRO: 0.23-a0-180-ga5275e7, CDO: 0.23-a0-178-g440b721, CJR: 0.23-a0-180-ga5275e7ff7, CPO: 0.23-a0-180-ga5275e7, CUE: 0.23-a0-180-ga5275e7ff7, CWZ: 0.23-a0-136-g722f094, CXC: 0.23-a0-180-ga5275e7, LLD: 0.23-a0-178-g440b721736 14:51:03 ok so CDO did rebuild 14:51:31 |amethyst: I wonder if windows trunk build on CDO is a separate repo that's caught in the same merge conflict 14:52:17 I wonder what's up with CWZ 15:11:44 03advil02 07* 0.23-a0-181-g24f899a: Revert complex bones install stuff, add dist_bones to repo 10(4 minutes ago, 37 files, 21+ 47-) 13https://github.com/crawl/crawl/commit/24f899ac9f8b 16:10:43 Unstable branch on crawl.jorgrun.rocks updated to: 0.23-a0-181-g24f899ac9f (34) 16:59:02 would it be difficult to make reflection kills award piety? 17:05:00 TIL if you use cause fear on an adjacent deep elf death mage it will cast vamp draining on you instead of running 17:06:07 <|amethyst> minmay: it would be mildly difficult to do so without also making them trigger god anger 17:06:15 <|amethyst> minmay: not impossible, I'm sure 17:06:32 <|amethyst> and probably not even all that difficult, but mildly so :) 17:07:22 it already works for reflected iron shots and the like 17:07:59 but not for reflected arrows, or a poison kill from a reflected poison attack, or a reflected Disintegrate from warlock's mirror 17:10:53 <|amethyst> hmm 17:11:18 <|amethyst> honestly that makes it sound more difficult :) 17:12:01 <|amethyst> wonder why iron shot would work but not arrows 17:12:13 <|amethyst> is it using the caster field? then why doesn't disint work? 17:12:40 advil: thanks, I fixed those win32 console zip filenames for 0.22.1 and 0.21.2 17:13:01 you ask that question like I would know the answer :P 17:17:22 <|amethyst> minmay: I was musing :P 18:21:55 Unstable branch on underhound.eu updated to: 0.23-a0-181-g24f899ac9f (34) 18:26:59 -!- neunon_ is now known as neunon 18:35:32 Blackbelt (L15 DsCK) ASSERT(you.equip[slot] == armour.link) in 'delay.cc' at line 772 failed. (D:13) 18:45:04 !crashlog 18:45:14 20043. Blackbelt, XL15 DsCK, T:20470 (milestone): https://underhound.eu/crawl/morgue/Blackbelt/crash-Blackbelt-20180918-223530.txt 18:45:58 heh 18:46:28 xom mutated the char causing horns 3 while they were taking their hat off, triggering a second armour off delay 18:56:16 I might leave that bug alone as a testament to xom 18:56:25 (also it'd be a pain to fix) 19:12:58 ??objstat[2 19:12:58 objstat[2/2]: Spreadsheets for all released versions and trunk: https://drive.google.com/folderview?id=0B7VXhHzhWWb7S282VWhLVWRXbG8&usp=sharing ; See the README for details: https://docs.google.com/document/d/1D5mFqVi8ghz_nzvVmDUc3unx8VanVBWfgvZ8xCHaiJo/edit?usp=sharing 19:23:39 duckroller: might be a silly question but are you running on a relatively modern PC which is not like a low power slow-clock-speed thing? 20:42:08 alexjurkiewicz: I've been testing on a range of 4-8 year old clusters, with mid-tier CPUs (Intel Xeon E5 2630 2.4 GHz, Xeon E5 5530 2.5 GHz, Xeon E5 2640 2.4 GHz). I'm generally parallelizing 1 crawl process per CPU core when I run my trials 20:44:11 also looking into reducing my search space to just trying to optimize D1 (likely in terms of death minimization, score/turncount seems silly), as those trials will take only a couple seconds each... But I'm still working out how to convince qw to quit the game after going down stairs 20:57:44 might be helpful http://doc.dcss.io/ 20:58:53 thanks 20:58:59 i'm not sure if lua actually exposes a "save and quit" function. You might need to script the entire process with something like expect (ugh) 21:00:37 actually, crawl.sendkeys(string.char(19)) 21:00:39 from https://github.com/crawl/crawl/blob/master/crawl-ref/source/test/stress/abyss.rc 21:06:30 Yeah I was trying to cop the quit code already used in qw for when it gets stuck. When stuck, it does this: 21:06:44 if stuck_turns > QUIT_TURNS then 21:06:48 magic(control('q') .. "yes\r") 21:06:53 return true 21:07:15 so it basically sends a ^q ... yes command over 21:08:53 which works, for qw, when it gets stuck. I'll look into the lua binding that abyss_walker uses 21:11:27 Yup, crawl.sendkeys(string.char(19)) worked a charm! Cheers! 21:29:05 -!- amalloy is now known as amalloy_ 21:51:02 -!- amalloy_ is now known as amalloy 22:39:47 -!- Tiobot is now known as Guest11145 23:59:45 Unstable branch on CRAWL.XTAHUA.COM updated to: 0.23-a0-181-g24f899a (34)