00:26:11 Crawl closes (crashes) immediately on startup 13https://crawl.develz.org/mantis/view.php?id=11872 by Siegurt 00:36:48 Crawl closes (crashes) immediately on startup 13https://crawl.develz.org/mantis/view.php?id=11873 by Siegurt 01:32:54 elliptic:have you noticed with qw that it fails to butcher if it's standing on two corpses, the top one being inedible and the lower one being edible? Or something like that 01:34:06 The wolf spider corpse isn't edible. Handle_corpses failed despite returning true. 01:34:13 that sounds plausible given that it used to be possible to butcher inedible corpses 01:34:44 (I haven't touched qw seriously in a couple versions) 01:40:51 one other question: do you have any system for comparing qw with and without a change? eg deciding if a change makes qw better or not 01:41:45 I've written a patch to use might potions more often, and I'm wondering how to evaluate it. The simplest way seems to be to run n seeded games without the patch, apply it, then run the same seeds again and compare the aggregate score. What do you think? 01:42:18 the big problem is you seem to need n of like ~100, and that takes a while to run 01:44:51 !lg salmon 01:44:53 3081. salmon the Carver (L10 VpFi of Kikubaaqudgha), demolished by a gargoyle (a +1 mace of holy wrath) (kmap: hangedman_cross_cluster) on D:10 on 2019-02-06 19:57:15, with 7485 points after 8670 turns and 0:17:33. 01:44:58 !lg salmon won 01:44:59 16. salmon the Messiah (L27 HOMo of Beogh), escaped with the Orb and 15 runes on 2018-11-01 11:35:39, with 16633927 points after 93320 turns and 5:04:25. 01:45:11 !lg * cwz won 01:45:12 6728. MeloName the Bringer of Law (L27 VSFi of Zin), escaped with the Orb and 15 runes on 2019-02-11 05:38:18, with 14810734 points after 107302 turns and 5:32:43. 01:45:20 !scorepage meloname 01:45:21 http://crawl.akrasiac.org/scoring/players/meloname 01:46:10 |amethyst: hong, the cwz admin, emailed me to say that cwz games appear to not be showing up in CAO scoring 01:46:11 that win from meloname is one example 01:46:13 !time 01:46:14 Time: Feb 11, 2019, 06:46:12 AM, UTC. The 2019 0.23 tournament ends in 13 days, 13 hours, 13 minutes and 47 seconds. 01:46:26 although it was quite recent, enough time to show up 01:46:34 !lg meloname won -2 01:46:35 4/5. MeloName the Conqueror (L27 DsFi of Kikubaaqudgha), escaped with the Orb and 15 runes on 2019-01-03 11:57:18, with 14281700 points after 112646 turns and 5:50:37. 01:46:43 !lg * cwz won -2 01:46:45 6727/6728. Yunor the Ruinous (L27 BaWn of Sif Muna), escaped with the Orb and 4 runes on 2019-02-11 03:39:53, with 2130688 points after 84889 turns and 5:54:14. 01:49:53 so here's a new pregen/seeding puzzle: til that the "acquire" itemspec literally calls the acquire code, and so it's therefore dependent on various aspects of the character, e.g. what armour they've seen, what their spell skills are 01:50:38 for pregen right now this means that deep "acquire" itemspecs won't generally be as good a fit, because they're all called on a blank slate 01:51:01 it also means that for non-pregen seeding hitting a level with an acquire spec will probably diverge from other seeds 01:51:09 er, from other instances of the same seed 01:51:36 not sure what the right thing to do is, but probably need to somehow remove the conditioning on the player 01:55:26 :O 01:55:44 you're talking about acquire as a kw for an item spec, right? 01:55:50 yeah, as in "acquire any" 01:55:51 s/kw/keyword/ 01:56:02 yeah; we'd sort of like to just phase those out 01:56:07 save possibly for trove 01:56:16 ok, remove acquire spec was an option I was actually considering 01:56:22 I feel a lot of maps use acquire to mean "put a randart here" 01:56:47 replace with "superb item" or whatever 01:56:51 yeah there's not too much thought put into it since many cases it was "acquire any" 01:57:03 which means it's likely to be useless 01:57:23 however I'm not actually sure what the remaining instances are 01:57:23 that would be my preferred option if vault people support that 01:57:34 there's a few misc ones that would need a closer look 01:57:59 sure, it could be superb item for acquire any; one of the main difference is that superb_item doesn't have consumables 01:58:10 so they could become e.g. any good_item 01:58:33 like 3 instances that use the god syntax ("acquire:sif_muna book") 01:58:36 but yeah even then, it's probably better to use superb_item and star_item most of the time 01:58:58 just to have the book be themed after sif? 01:59:12 that's cosmetic only, right? 02:00:04 not sure, it's in a trove so it's probably some special case 02:00:09 haven't looked at the map yet 02:00:19 book acq in general does a bunch of conditioning on skills 02:00:35 yeah, and that same code is used for e.g. sif gifts 02:02:06 but it's probably fine to remove acquire keyword if we can figure out what to do for troves 02:02:34 huh is "any good_item" better than "superb_item"? 02:02:52 that's what I might conclude from the docs, but I'm not sure how to read this 02:03:17 oh maybe they're similar but with different item sets? 02:07:00 for my purposes making it deprecated but allowed in trove would make sense 02:07:34 maybe also a few cases in sprints where it's needed to generate something on the fly 02:07:51 since trove is not pregenerated anyways 02:15:42 advil: any covers all item categories, superb_item is restricted to equipment type items 02:15:45 ??superb_item 02:15:45 superb item[1/1]: In {des} syntax, an item with the same quality as {good_item} but with the following type dist.: 2/7 weapon, 1/7 armour, 1/7 jewellery, 1/7 book, 1/7 misc, 9/70 staff, 1/70 rod. The | glyph is defined as this by default. 02:16:28 so for loot we often use | (superb_item) and * (star_item), the latter can include things like gold, potions, scrolls 02:17:02 star_item has a depth-based quality that isn't good_item quality 02:17:13 ??star item 02:17:14 star item[1/1]: In {des} syntax, an item of quality (5 + depths * 2). The * glyph is defined as this by default. 02:28:41 -!- amalloy is now known as amalloy_ 02:39:19 New branch created: pull/986 (1 commit) 13https://github.com/crawl/crawl/pull/986 02:39:19 03NormalPerson702 07https://github.com/crawl/crawl/pull/986 * 0.24-a0-68-g4e72f50: Fix yellow draconians' rAcid mutation not being applied (malacante) 10(17 minutes ago, 1 file, 1+ 0-) 13https://github.com/crawl/crawl/commit/4e72f50126cd 02:49:03 Monster database of master branch on crawl.develz.org updated to: 0.24-a0-93-ge1bfdf8 02:55:23 -!- redhound is now known as brittlin 03:10:19 -!- brittlin is now known as redhound 03:11:16 -!- redhound is now known as brittlin 03:30:53 Fork (bcrawl) on crawl.kelbi.org updated to: 0.23-a0-1682-g99b902e015 05:13:59 !tell gammafunk maybe klown pie effects should go into the morgue's note log in the same way mutations do 05:14:01 alexjurkiewicz: OK, I'll let gammafunk know. 07:18:03 advil: please include the seed in crash textfile output near the top. It says "Game mode: seeded" but not the actual seed 09:24:36 Stable (0.22) branch on underhound.eu updated to: 0.22.1-7-g5c428e77e6 09:39:53 alexjurkiewicz it's not in online crashdumps becaues it's such a big information leak, should be there in offline 09:40:45 where online = USE_DGAMELAUNCH build 10:51:20 03advil02 07* 0.24-a0-94-g13aa1b5: Clear even more state for tests 10(32 minutes ago, 1 file, 36+ 8-) 13https://github.com/crawl/crawl/commit/13aa1b5c4e15 10:51:20 03advil02 07* 0.24-a0-95-g4f8aab8: Add a debugging hook 10(28 minutes ago, 2 files, 11+ 1-) 13https://github.com/crawl/crawl/commit/4f8aab8ed993 10:51:20 03advil02 07* 0.24-a0-96-g88f18ea: constify a bunch of instances of PlaceInfo 10(24 minutes ago, 2 files, 56+ 48-) 13https://github.com/crawl/crawl/commit/88f18ea6ed1b 10:51:20 03advil02 07* 0.24-a0-97-g60e8692: Deprecate acquire item specs for regular dungeon generation 10(13 minutes ago, 13 files, 39+ 37-) 13https://github.com/crawl/crawl/commit/60e86920ebe8 10:51:20 03advil02 07* 0.24-a0-98-g1c07cc1: Small tweak to vault_catalog.lua 10(50 seconds ago, 1 file, 1+ 0-) 13https://github.com/crawl/crawl/commit/1c07cc13e6d9 10:58:53 alexjurkiewicz: yeah, you need a pretty large n to have any sort of certainty on whether most changes help (if the change helps more, you need smaller n to be confident that it's a positive change) 10:59:36 it is one of the reasons why I've always been interested in optimizations to make crawl run faster 11:01:02 one thing that does help if you aren't already doing it is to run multiple games at once (up to the number of cpu threads you have) 11:15:24 03advil02 07* 0.24-a0-99-ge52b27f: Place AK start in abyss with pregeneration on (11871) 10(2 minutes ago, 2 files, 13+ 9-) 13https://github.com/crawl/crawl/commit/e52b27f00579 11:16:46 Unstable branch on crawl.kelbi.org updated to: 0.24-a0-98-g1c07cc13e6 (34) 11:32:38 Unstable branch on crawl.kelbi.org updated to: 0.24-a0-99-ge52b27f005 (34) 12:04:15 The build was fixed. (master - 1c07cc1 #11120 : advil): https://travis-ci.org/crawl/crawl/builds/491698174 12:11:53 -!- amalloy_ is now known as amalloy 12:19:39 Unstable branch on crawl.akrasiac.org updated to: 0.24-a0-99-ge52b27f (34) 13:11:33 The build was fixed. (master - e52b27f #11121 : advil): https://travis-ci.org/crawl/crawl/builds/491709950 15:41:19 New branch created: pull/987 (1 commit) 13https://github.com/crawl/crawl/pull/987 15:41:20 03alexjurkiewicz02 07https://github.com/crawl/crawl/pull/987 * 0.24-a0-94-gf5cc8d7: Run GDB from /usr/local/bin on OSX 10(10 days ago, 3 files, 9+ 2-) 13https://github.com/crawl/crawl/commit/f5cc8d7df097 16:35:33 the new console clouds are very bubbly 16:36:03 yeah 16:36:21 some of the glyphs look too similar to crawl monsters 16:36:28 well the one that's like an asterisk 16:36:34 but I haven't customized mine yet 16:37:01 I was definitely thinking about glyph overhauls recently (another side effect of playing some brogue), but not yet sure about this one as a default 16:44:37 it was hard to find four glyphs in the intersection of codepages we've decided to restrict to that looked right and weren't taken 16:45:11 I'm not complaining too hard, it's not too bad :) 16:45:25 why do we have the codepage restriction? 16:45:44 no idea! it's what was in the comments and I didn't want to rock the boat 16:48:26 emoji monsters when 16:49:30 alexjurkiewicz: 🗑 16:50:01 I think codepage restriction is for supporting older terminals 16:51:08 I think it would be fine to let people with old terminals configure their glyphs themselves, we could provide a config set 16:51:28 glyph reform 0.24 16:51:42 for the whole 5% of the player base that uses console 16:52:46 crawl handles unicode fine, though I'm not sure what fonts guarantee what unicode glyphs to be single width 16:52:52 and if that's even queryable 16:53:36 ebering: the moment you mention width of glyphs in fonts you are embarking on some very deep magic, as i understand it 17:00:29 advil: 5% of the player base, 75% of the devs :D 17:00:43 you are embarking upon a fool's errand, since crawl in console mode can't tell what font the terminal is using 17:01:08 I would be keen on a -headless option btw, which only prints the message log (or perhaps nothing at all) for testing qw with 17:01:08 nor ask it for font or character metrics 17:01:41 well, I think what we'd do is target a few specific modern default setups 17:01:43 alexjurkiewicz what's wrong with fake_pty? 17:03:32 if you want to see the message log while using fake_pty, you may be interested in a recent commit of mine 17:03:36 %git 4f8aab8 17:03:36 07advil02 * 0.24-a0-95-g4f8aab8: Add a debugging hook 10(7 hours ago, 2 files, 11+ 1-) 13https://github.com/crawl/crawl/commit/4f8aab8ed993 17:04:10 you need to recompile to use it but just add set_log_emergency_stderr(true); somewhere towards the beginning of things 17:07:21 advil: nothing is wrong with it except I am looking to make crawl as fast as possible as per elliptic above 17:07:47 not figuring out how to draw the map seems like a potential improvement 17:11:26 I would be that the actual drawing isn't as involved as world_reacts 17:11:46 *bet 17:12:31 and that most of the drawing time is wasted with actual drawing that is mitigated already by fake_pty (though that's more of a guess, just based on interacting with fake_pty a lot) 17:12:52 but if you run fireworks.rc or something, it's already *way* faster under fake_pty 17:15:57 most of the delay there comes from the time the terminal takes to render, and the limited size of the buffer meaning crawl blocks a lot on output. fake_pty isn't rendering etc. and just discards what it gets, so can keep up with the stream 17:16:12 there will be some slowdown from crawl's drawing, but not that much 17:18:17 is it an intentional change that slime can require walking next to slimewalls? 17:18:31 there are vaults that do that, yeah 17:21:54 thanks :( 17:23:00 rchandra: you can dig those walls through, and of course rCorr is one resistance potion away 17:24:58 I chose to save my ?ID for a trove, and will die in here with 2 runes thanks to qaz not being at full power 17:25:15 thanks for nothin qaz 17:36:40 there needs to be a feature where after I come to the conclusion that what I'm doing is bad, it makes me stop doing it 17:36:57 rchandra: There is, it's "dying" :-/ 17:37:10 in particular, I didn't need to have like 8 near-deaths in slime before dying, instread of leaving 17:37:15 fair 17:37:48 (no MR slime is bad, can't Qaz 100% of golden eyes) 18:05:50 rchandra: well what I'd probably try to do is keep some elementals up most of the time 18:05:57 they're not too expensive and they can't be confused 18:06:40 obviously you can upheaval in most cases but just creating some allies that can't get confused is pretty handy 18:24:36 Unstable branch on underhound.eu updated to: 0.24-a0-99-ge52b27f005 (34) 18:40:52 ready to join the official clan 18:43:20 do you have a 200% winrate 18:43:50 its a percentage of that, yeah 18:44:00 a fraction if you will 18:44:19 its like 1/3 of those numbers you listed 18:44:42 look, and only for a moment: i try hard at these things 18:46:37 -!- TigerbotHesh_ is now known as TigerbotHesh 18:48:09 gammafunk we can start an elite clan 18:48:39 it will spawn a new monster type: phasing acidhydras 18:49:14 lernean acid bite 18:49:25 something pure ownage like that 19:32:12 thanks gammfunk I'm not going to give up on homo/qaz/axe/15 for a few more splats yet 20:09:49 alexjurkiewicz: how are you installing gdb on os x? 20:10:28 afaik there's no apple distributed version so no standardized path at all 20:11:23 maybe that's where homebrew would install it? not macports or anaconda though 20:12:28 wouldn't it be wiser to use a POSIX utility to find the location of gdb? 20:17:05 probably 20:17:11 we hardcode the path for linux already 20:17:22 also would be good to try lldb too 20:17:28 something I've never gotten around to adding 20:31:06 so this is slightly insane but it appears that the constructor for you (maybe other globals) is not getting called on mingw builds 20:32:25 anyone know if something like this has come up before? 20:33:07 that's the reason for the current crash on trunk builds, and also probably the place_info save corruption 20:33:43 not entirely sure what changed between 0.23 and the trunk builds, but probably something fairly indirect 20:35:25 do we have another ordering issue? 20:35:52 (globals don't get constructed in any specific order.) 20:35:54 afaict it should be ok wrt global ordering, but there's some weird stuff in there to do with sdl that's giving me pause 20:36:17 the crucial globals are in the compilation unit with main so the spec says they should get constructed (in some order) before main is called 20:39:35 advil: how did you conclude that it didn't get constructed? 20:51:25 mm, I'm not entirely sure it didn't get constructed, but by looking at the object in gdb, some things that are supposed to have happened in the constructor didn't 20:51:44 also I manually initialized you via assignment in main() and this crash went away 20:52:17 the crash is when trying to get a (previously unsed) value in you.props, where if you.props is setup right, should create the value 20:52:36 I can also see that you.branch_info isn't initialized right 20:56:46 you mentioned this is mingw; are you testing via msys2? 20:57:29 yes 20:57:49 but the cross-compiled build is broken too, that's why I was testing 21:10:56 hm new hypothesis, it is *other* things that are not being constructed 21:10:56 so it maybe is a global order issue 21:11:13 at the time of the player constructor call, branches is just a list filled with BRANCH_DUNGEON 21:11:28 not sure yet about the props crash 21:18:09 -!- amalloy is now known as amalloy_ 21:31:48 yeah, both of these are globals ordering thing, the current crash is because a static const string is empty at the time that the constructor for `you` is called 21:33:46 -!- amalloy_ is now known as amalloy 21:48:30 advil: with homebrew 21:48:55 advil: I don't mind creating a symlink myself, and this seemed like the lowest complexity change to get gdb at least theoretically working 21:49:15 like, however you install gdb, you can symlink into /usr/local/bin 22:19:29 alexjurkiewicz: btw I don't really see noting klown pie effects; they're all non-permanent, unlike mutations. It would be like noting everying time a lich hexed you, or something 22:20:00 and, furthermore, consider the following 22:20:00 !klownpie alexjurkiewicz 22:20:00 The Killer Klown throws a pie at alexjurkiewicz. The Klown pie hits alexjurkiewicz! Raspberry! alexjurkiewicz feels weakened. alexjurkiewicz feels dopey. alexjurkiewicz feels clumsy. 22:20:15 !klownpie advil 22:20:15 The Killer Klown throws a pie at advil. The Klown pie hits advil! Lemon! advil is unable to drink. 22:30:38 your arguments (both of them) are persuasive 22:30:52 I guess you can already tell what klown pie effects are most deadly because they apply a status effect which shows up in !lg x=status 22:31:36 yeah 22:31:39 !cmd .klowned 22:31:40 Command: .klowned => !lg * newklowns (( cikiller=killer_klown || zot status~~bat-form|wisp|pig-form|fire_vuln|silence|vertigo|no_potions || zot collapse|brainless|clumsy )) title:"Killed by new Killer Klowns$(argst $*)" $* 22:31:47 is what I've been using to find klown deaths 22:31:51 .klowned 22:31:54 65. gjx2 the Vine Barricade (L27 VSFi of Zin), slain by a storm dragon on Zot:5 (hall_of_Zot) on 2019-02-12 01:39:47, with 1349745 points after 92206 turns and 2:43:28. 22:32:10 it has false positives in various ways but it works fairly well 22:33:31 I wonder if when loading saves, the game should do `you = player()` like it apparently does for new games 22:34:48 (no wonder it's taken me forever to track this stuff down) 22:35:21 advil: i think it does, via _reset_game? 22:35:31 not if you load a game right off 22:36:56 but on first load, you gets set to player() automatically when the global is loaded, no? 22:37:35 i guess your point is that the side effects of player() will be different when the global is loaded, because it depends on a bunch of other stuff? 22:37:39 right, but the problem is that it's a bit finicky about what you can put in there because it is a global 22:38:07 I'm guessing this is why the new game call is there because someone partially fixed a bug like this 22:49:52 03advil02 07* 0.24-a0-100-g3339647: Simplify levels_visited 10(60 minutes ago, 3 files, 0+ 12-) 13https://github.com/crawl/crawl/commit/33396473a155 22:49:52 03advil02 07* 0.24-a0-101-ge13aadb: Eliminate some more dependencies on globals from player::player 10(20 minutes ago, 2 files, 17+ 5-) 13https://github.com/crawl/crawl/commit/e13aadb210ac 22:52:48 -!- muffindrake1 is now known as muffindrake 23:05:44 Reduction in wand charges sits uneasily with digging trove. 13https://crawl.develz.org/mantis/view.php?id=11874 by damerell 23:13:34 uh hm 23:14:11 03advil02 07* 0.24-a0-102-g7635c99: Revert "Eliminate some more dependencies on globals from player::player" 10(25 seconds ago, 2 files, 5+ 17-) 13https://github.com/crawl/crawl/commit/7635c99b3565 23:14:25 gammafunk: looks like silence and no potions are the worst, based on looking at # of deaths and seeing which statuses have the most 23:15:20 alexjurkiewicz: I wouldn't try to reason about what is actually the worst based on sequell and based on such a small sample size 23:15:28 annoyingly, that commit breaks a rare save compat fix in a fairly bad way 23:15:34 also -potions has false positives with ru, as does silence potentially 23:15:36 with zot traps 23:15:59 I'm glad I reverted before anything updated 23:16:30 if you look at sequell and filter out all the false positives you can see what players tend to die with but that's not a good indicator of what's truly dangerous 23:16:39 another thing is that polymorph has lowest weight 23:16:48 and rF- is also lower weight (not as low as poly) 23:16:58 so you'd have to account for that as well 23:22:32 woah, player::player()...ah that's the constructor 23:22:57 i imagine a lot of players die while silenced in zot because they cast the silence spell or used a scroll of silence 23:25:02 weird CAO issue: the most recent streak for a player (http://crawl.akrasiac.org/scoring/players/sauken.html), consisting of games from jan/feb is described as being broken by a CJR game from months ago 23:26:29 Unstable branch on crawl.kelbi.org updated to: 0.24-a0-101-ge13aadb210 (34) 23:26:35 |amethyst I guess it is you that should be pinged about ^ ? 23:32:52 Unstable branch on crawl.kelbi.org updated to: 0.24-a0-102-g7635c99b35 (34) 23:43:44 oh, the other issue is that cjr morgue links need to go to the S3 url 23:43:48 !lg . cjr -log 23:43:49 70. gammafunk, XL27 GrSk, T:86876: https://s3-us-west-2.amazonaws.com/crawl.jorgrun.rocks/morgue/gammafunk/morgue-gammafunk-20180826-072239.txt 23:47:08 yeah, that too 23:47:23 !lg sauken cjr opfe -log 23:47:24 1. Sauken, XL1 OpFE, T:44: https://s3-us-west-2.amazonaws.com/crawl.jorgrun.rocks/morgue/Sauken/morgue-Sauken-20180720-191208.txt 23:47:44 that's the supposed streak-breaker 23:48:11 where did the cjr milestone file end up? 23:48:15 files 23:49:33 <|amethyst> hrm 23:49:52 <|amethyst> probably needs a database rebuild, though I don't know what that will do about streaks 23:50:15 <|amethyst> those cjr games are probably duplicated, too 23:50:24 |amethyst: hong (cwz admin) also contacted me to say that cwz games aren't showing up in CAO scoring since... 23:50:43 he said jan 27th, but I saw some from a couple days later 23:50:53 but definitely seemed to be missing cwz 23:50:56 I wonder what can be done about cao scoring 23:51:04 it's just getting creakier and creakier 23:51:10 <|amethyst> yeah 23:51:39 would rule if sequell could generate the pages 23:51:39 <|amethyst> and I don't really have time to work on it other than occasionally restarting it and tweaking the sources