00:11:26 if you want to modify the action you could vendor it or submit a PR. I haven't submitted it to the marketplace yet because I figured someone would raise an ownership concern 01:38:49 Unstable branch on crawl.develz.org updated to: 0.26-a0-146-g24f6d3962c (34) 03:31:59 Fork (bcrawl) on crawl.kelbi.org updated to: 0.23-a0-3303-gb091788534 03:49:27 Are 0.22 and earlier tournament pages availalble somewhere? 04:03:43 /join ##crawl 05:35:33 -!- amalloy is now known as amalloy_ 08:18:52 New branch created: include-what-you-use (1 commit) 13https://github.com/crawl/crawl/tree/include-what-you-use 08:18:52 03Aidan Holm02 07[include-what-you-use] * 0.26-a0-136-g5414c13: Move a bunch of preprocessor macros out of AppHdr.h 10(7 minutes ago, 2 files, 75+ 65-) 13https://github.com/crawl/crawl/commit/5414c130202e 08:27:17 I like the code support we have for solaris 08:30:16 04Build failed for 08include-what-you-use @ 5414c130 06https://github.com/crawl/crawl/actions/runs/151407769 08:32:34 there's a bunch of really weird stuff lying around if you look in the right places 08:55:46 !kw nchoice 08:55:46 Keyword: nchoice => MfEn|BaEE|HaVM|HaFE|CeIE|CeWz|TeSk|OpAE|FoNe|GhCj|BaAK|DDAM|HuTm|SpAK|DEAs|OgCj|VpWr|KoIE|VSEE|DgEn|OpHu|GnAE|VSHu|KoSk|GhSu|TeTm|DEGl|SpFE|MuAK|VpAM|MfNe|KoWz|FoWr|GrEn|BaCK|HOCj|VSVM|FeAK|HaWz|CeEE|GhEE|NaAM|MuSk|DDIE|DgHu|OgEn|SpSk|TeAs|HOAE|DDCj|CeSu|TeWr|HaIE|KoNe|VpHu|KoVM|VSAE|HuAM|FoTm 09:23:15 The build passed. (include-what-you-use - 5414c13 #13692 : Aidan Holm): https://travis-ci.org/crawl/crawl/builds/703121374 10:00:22 gammafunk: btw since you're using vim, you might find vim-lsp useful 10:00:53 i've found ctags tends to be kinda stupid and takes me deep into random python files 10:02:24 aidanh: I clicked through three links from there via "language server protocol" and I still don't know what language server protocol is! 10:02:31 one of which took me to a microsoft website 10:02:46 oh here we go: https://microsoft.github.io//language-server-protocol/overviews/lsp/overview/ 10:02:58 it's a protocol for your editor to talk to a daemon which handles all the actual stuff 10:03:45 this works well in practice for you? 10:03:45 I've had the ctags/python problem as well 10:04:09 yeah it's pretty nice, although it has trouble with crawl's header files 10:04:15 but the prediction is better and faster 10:05:08 i basically copy-pasted the stuff on https://jonasdevlieghere.com/vim-lsp-clangd/, and for the compile_commands.json thing i'm using https://github.com/rizsotto/Bear 10:05:45 it also integrates with autocomplete; took me maybe 15 minutes to get it working and add some nice keybinds 10:05:57 oh, autoformatting with clang-format too 10:06:06 03advil02 07* 0.26-a0-147-g644bcb5: Seed stability fixes for heliophobic_arrival_battle_scene 10(16 minutes ago, 3 files, 6+ 1-) 13https://github.com/crawl/crawl/commit/644bcb522d4a 10:08:22 aidanh: thanks, I will look into these 10:16:05 report that some clouds are invisible in trunk 10:16:42 at least poison gas in particular 10:16:53 Unstable branch on crawl.kelbi.org updated to: 0.26-a0-147-g644bcb522d (34) 10:20:10 yep, can reproduce 10:21:03 that is probably a rendering change i did recently; i'll take a look 10:28:45 03advil02 07[stone_soup-0.25] * 0.25.0-16-g6b82105: 0.25-specific seed fix for heliophobic_arrival_battle_scene 10(8 minutes ago, 1 file, 7+ 9-) 13https://github.com/crawl/crawl/commit/6b821058a4fb 10:31:04 got to love all those weird des seed stability issues 10:32:15 yeah corpse generation is seriously wtf 10:32:35 I think it does make sense in the end because it allows the use of arbitrary monspecs to generate the corpse 10:32:46 perhaps giving each vault a nested rng generator would help mitigate this more decisively 10:32:55 maybe 10:33:22 there's a tradeoff there, in terms of rng space, and I have no sense for whether it's an impactful tradeoff 10:33:38 rng space? 10:33:50 as in the space of possible outputs of the rng? 10:33:53 yeah 10:34:36 everything I isolate is determined by a single roll from the current generator 10:34:36 i would think it would have zero impact, unless the rng output is somehow predictable, which sounds like a bigger issue 10:35:21 not impact in terms of players being able to predict, but impact in terms of how much of the possible space in principle the system is using 10:36:34 I'm also not sure that isolating the vault would've helped in this case 10:36:47 since the details of the vault interact in chaotic ways with everything else on the level 10:37:22 e.g. in different species cases I looked at the vault was rotated differently or even vetoed 10:38:12 right, there is that aspect as well 10:38:40 it would've dealt with the obvious issue in the vault lua probably 10:39:16 though I guess even that depends on when rotation is handled relative to the subgenerator 10:40:06 anyways this sort of reasoning is why I have overall tried not to spend *too* much time outthinking the rng in advance 10:40:29 what I'm more interested in are ways to try to keep the overall choices as stable as possible in the face of things like changes to the vault list 10:44:19 it's possible to have some seed stability even if the vault list changes? 10:44:31 good question! 10:44:37 but we currently don't have that 10:45:35 the space issue is this: suppose 10 64 bit rolls are done in a subgenerator (which is an abstraction, because it is extremely rare to use all 64 bits in practice). The subgenerator state is determined by a single 64bit roll (or some constant number of rolls). So the space of possibilities that you would have if just doing those rolls directly on the main generator is vastly huger than you get in the subgenerator. But like I said, the 10:45:35 impacts of this if any are .. hard to understand 10:47:09 i don't have a convincing argument that the state space impact is negligible, and i'm not 100% certain it is, but i have a very strong intuition that it's probably not worth worrying about 10:49:49 I also suspect it may not matter? 10:50:00 and I'm already doing a lot of subgeneratoring 10:50:06 e.g. each branch is in its own 10:50:12 I may eventually move each level to its own 10:51:08 rng math is such a mysterious art though and there are so many counterintuitive/hard-to-understand results that it does still make me nervous 10:52:25 New branch created: fix-invisible-poison-clouds (1 commit) 13https://github.com/crawl/crawl/tree/fix-invisible-poison-clouds 10:52:25 03Aidan Holm02 07[fix-invisible-poison-clouds] * 0.26-a0-136-g6a48f1e: Fix clouds being invisible 10(2 minutes ago, 2 files, 4+ 2-) 13https://github.com/crawl/crawl/commit/6a48f1e94a48 10:52:50 it'd be nice if we had some statistical tests 10:56:15 gonna let that go through the CI then merge 10:56:37 not sure if that assert will trigger (although it really shouldn't...) 10:57:04 haha write "// should be safe" and it'll be fine 10:58:20 the magic words 10:59:12 has that one ever been posted on the twitter? Anyone want to give me access to @crawlcode? 10:59:22 re keeping things stable modulo changes to the vault list, isn't that impossible? 10:59:57 it's impossible the way I described it, but there are degrees of stability 11:00:06 since you mention @crawlcode, i just recently tried to chart crawl's header file deps: https://0x0.st/iJJI.png 11:01:11 it occurs to me that one way to avoid instability after removing vaults is to just not remove them 11:01:13 for example, here's a scheme that does it: each version, fix a vault list as the "stable" list and then for any vaults not on that least, evenly pull a tiny bit of probability according to the appropriate weights from the stable list and distribute it across everything on the new list 11:01:37 for removals, you evenly give probability back across the whole remaining list 11:01:52 (this is exceedingly difficult to implement) 11:02:11 ah, so instability across versions is fine? 11:02:15 cerealjynx (L18 TrFi) ERROR: range check error (20 / 20) (D (Sprint)) 11:02:33 !crashlog 11:02:34 22289. P0WERM0DE, XL18 DsCK, T:38964 (milestone): https://underhound.eu/crawl/morgue/P0WERM0DE/crash-P0WERM0DE-20200628-172754.txt 11:02:35 yeah, that scheme requires marking some checkpoints to try to remain faithful to 11:02:43 !crashlog 11:02:44 22289. P0WERM0DE, XL18 DsCK, T:38964 (milestone): https://underhound.eu/crawl/morgue/P0WERM0DE/crash-P0WERM0DE-20200628-172754.txt 11:02:57 urgh, i could never get the hang of sequel 11:03:02 it's a sprint game 11:03:08 !crashlog cerealjynx sprint 11:03:09 1. cerealjynx, XL18 TrFi, T:1925 (milestone): http://crawl.akrasiac.org/rawdata/cerealjynx/crash-cerealjynx-20200629-150215.txt 11:03:50 things with the sprint keyword never show up unless it is explicitly there 11:03:50 ah, thanks 11:03:55 looks like another render bug 11:08:41 there are other simpler things that would help too, for example not using the full vault list in a lot of cases (e.g. right now I think overflow altar vault selection may be impacted by the size of the overall vault list, though I haven't confirmed that) 11:08:45 that sounds like the intent is to generate exactly the same map, but swapping removed vaults for new vaults (or existing ones) 11:11:21 so one possibility is to do that; don't ever remove vault definitions, and still use them for initial generation, but as soon as each level finished generating, try to swap them with any available vault 11:11:33 maybe...that sounds pretty hard, but it's more like: right now if you would draw vaults B, C, D at some point in seed x and vault A gets removed, at that point in seed x you may well get vaults N, Q, Z or whatever 11:11:57 right, makes sense 11:11:58 because choosing randomly from a list is impacted in very complex ways by any changes to list size 11:12:16 not deleting vaults like you suggest would also help 11:12:32 dealing with added vaults is tricky 11:13:49 reminds me of an article i read about scheduling oncall rotations, and how changing one entry would cause entire schedules to change 11:15:07 yeah, with this approach you'd basically have to fix on the current set of vaults, and newer vaults would have to be swapped in later 11:15:30 but that could significantly reduce their likelihood of being selected if they have complex rules / vetoes / etc 11:17:49 yeah 11:20:19 what about encoding a version number into the seed, and then using that to determine the set of vaults to use? 11:20:29 ah interesting...encoding version number in the seed is definitely on my list but I hadn't actually thought of the second part, was just going to use it to give a warning 11:20:46 there's a maintenance burden because you'd also need to record changes 11:21:05 record changes to what; vaults? 11:21:07 yeah 11:21:54 well, the seeds i've seen are kind of large, i'd say there's enough space to just bump the version numbers every time you do so 11:22:24 ah, you mean you'd need to store each version of each vault 11:22:26 yeah 11:22:44 hm, yes, that's a fly in the ointment alright 11:31:52 I wonder what the best way to handle a post-tourney debrief is? ebering is on vacation now right? 11:32:01 wondering partly because I can tell people want to talk about it 11:32:19 advil: well, I can make one, but I'd sort of thought ebering might want to 11:32:24 if he's away I can do 11:32:24 yeah 11:32:39 it would make sense to just wait maybe 11:32:45 I'm fine with that 11:33:51 we've never been too prompt about the summary, and I'd done very reduced summaries in recent years due to trolling and general hostility from a certain group of players 11:34:26 but since ebering was so instrumental in the new tournament design, maybe he'd like a chance to make a more extensive summary 11:58:48 The build passed. (fix-invisible-poison-clouds - 6a48f1e #13695 : Aidan Holm): https://travis-ci.org/crawl/crawl/builds/703188159 12:12:20 03Aidan Holm02 07* 0.26-a0-148-g5ef36ef: Fix clouds being invisible 10(82 minutes ago, 2 files, 4+ 2-) 13https://github.com/crawl/crawl/commit/5ef36ef8a83d 12:16:33 Unstable branch on crawl.kelbi.org updated to: 0.26-a0-148-g5ef36ef8a8 (34) 12:29:12 does that need rebuilds? 12:29:48 might be a good idea 12:33:50 Unstable branch on crawl.akrasiac.org updated to: 0.26-a0-148-g5ef36ef (34) 12:34:48 thanks, i see you've already got cbro 13:01:29 Stable (0.25) branch on crawl.akrasiac.org updated to: 0.25.0-16-g6b82105 14:17:23 ??epic bugs[1 14:17:23 epic bugs[1/28]: if (you.religion = GOD_XOM) 14:19:09 !learn add epic_bugs if (crawl_state.type = GAME_TYPE_CUSTOM_SEED) (see also, epic_bugs[1]) 14:19:09 epic bugs[29/29]: if (crawl_state.type = GAME_TYPE_CUSTOM_SEED) (see also, epic_bugs[1]) 14:46:58 man, I just cannot keep my bug stack at size 1 the last few days 14:51:01 03gammafunk02 07* 0.26-a0-149-g49b1be5: Fix loot in nicolae_gauntlet_exit_dead_adventurers 10(48 minutes ago, 1 file, 7+ 7-) 13https://github.com/crawl/crawl/commit/49b1be5409ea 14:53:49 03gammafunk02 07[stone_soup-0.25] * 0.25.0-17-g773c8c6: Fix loot in nicolae_gauntlet_exit_dead_adventurers 10(51 minutes ago, 1 file, 7+ 7-) 13https://github.com/crawl/crawl/commit/773c8c618939 15:01:29 Unstable branch on crawl.kelbi.org updated to: 0.26-a0-149-g49b1be5409 (34) 15:01:57 advil and I had a partial discussion about this, but fr: des build errors like the one I caused in commits above get checked via some part of CI 15:02:09 or I guess ideally they'd just cause a fatal error 15:02:45 apparently because I'm passing loot def into a function call it couldn't be caught so easilly 15:05:16 it probably can be caught, esp since it sounded from what you said like the error was in the log 16:21:52 New branch created: pull/1470 (1 commit) 13https://github.com/crawl/crawl/pull/1470 16:21:52 03mainiacjoe02 07https://github.com/crawl/crawl/pull/1470 * 0.26-a0-150-gbdd78ce: Give Artificers a choice of starting weapon 10(12 minutes ago, 1 file, 2+ 2-) 13https://github.com/crawl/crawl/commit/bdd78ce26e29 18:22:38 Unstable branch on underhound.eu updated to: 0.26-a0-149-g49b1be5409 (34) 18:53:05 does trunk still have the tournament reminder? 20:17:05 03kate-02 07* 0.26-a0-150-g420014e: Revert "Add a trunk reminder for the 0.25 tournament" 10(4 hours ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/420014ea0472 20:17:42 alexjurk1: nope! 20:17:54 (thanks for the reminder) 20:22:23 Stable (0.25) branch on underhound.eu updated to: 0.25.0-17-g773c8c6189 20:31:30 Unstable branch on crawl.kelbi.org updated to: 0.26-a0-150-g420014ea04 (34) 21:40:15 03advil02 07* 0.26-a0-151-g4a859bd: Fix epilogue reloading 10(4 hours ago, 1 file, 12+ 3-) 13https://github.com/crawl/crawl/commit/4a859bdc29e1 21:40:15 03advil02 07* 0.26-a0-152-gb10996a: Fix a format specifier 10(2 hours ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/b10996a73cb5 21:40:15 03advil02 07* 0.26-a0-153-gdbaac09: Use last game type for startup menu, cleanup 10(23 minutes ago, 1 file, 14+ 12-) 13https://github.com/crawl/crawl/commit/dbaac098b9f9 21:40:15 03advil02 07* 0.26-a0-154-g3d63440: Don't let hints global state persist across restarts 10(5 minutes ago, 1 file, 2+ 0-) 13https://github.com/crawl/crawl/commit/3d63440f564c 21:43:29 03advil02 07[stone_soup-0.25] * 0.25.0-18-g0f731e5: Optimize some default autopickup lua code 10(2 days ago, 1 file, 5+ 2-) 13https://github.com/crawl/crawl/commit/0f731e504686 21:43:29 03Goratrix02 {advil} 07[stone_soup-0.25] * 0.25.0-19-g4a33e3d: Fix scarf of harm not working 10(7 days ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/4a33e3d59c9a 21:43:29 03Gittourarmy02 {advil} 07[stone_soup-0.25] * 0.25.0-20-gee69a8b: Show full name for orc resurrection prompt 10(4 weeks ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/ee69a8bff294 21:43:29 03Elliot Dronebarger02 {advil} 07[stone_soup-0.25] * 0.25.0-21-ga8ac842: Fix segfaults when casting spells from wizmode (11889) 10(4 weeks ago, 1 file, 8+ 2-) 13https://github.com/crawl/crawl/commit/a8ac842e593d 21:43:29 03advil02 07[stone_soup-0.25] * 0.25.0-22-gc55729f: Optimizations for drawing many gold piles 10(34 hours ago, 4 files, 20+ 4-) 13https://github.com/crawl/crawl/commit/c55729fc0252 21:43:29 03PleasingFungus02 {advil} 07[stone_soup-0.25] * 0.25.0-23-g0cd25f8: Mark !stab useless under TSO (Yermak) 10(10 days ago, 1 file, 2+ 0-) 13https://github.com/crawl/crawl/commit/0cd25f802a72 21:43:29 03advil02 07[stone_soup-0.25] * 0.25.0-24-g1780501: Fix epilogue reloading 10(5 hours ago, 1 file, 12+ 3-) 13https://github.com/crawl/crawl/commit/1780501fa57f 21:43:29 03advil02 07[stone_soup-0.25] * 0.25.0-25-gc2ca73c: Use last game type for startup menu, cleanup 10(26 minutes ago, 1 file, 14+ 12-) 13https://github.com/crawl/crawl/commit/c2ca73c7a94a 21:43:29 03advil02 07[stone_soup-0.25] * 0.25.0-26-g0c896c7: Don't let hints global state persist across restarts 10(8 minutes ago, 1 file, 2+ 0-) 13https://github.com/crawl/crawl/commit/0c896c7088f9 21:46:50 Unstable branch on crawl.kelbi.org updated to: 0.26-a0-154-g3d63440f56 (34) 22:07:01 please ping the @serveroperators group when 0.25.1 comes out 22:09:50 oh, good idea 22:10:00 I still have one more bug to fix 22:16:02 There's a weird interaction when you have DChan up and you are using WJC's whirlwind ability and you kill a monster with your first few swings in a move and still have attacks left. it asks you if you really want to attack your spectral thing, but putting N doesn't make the dialog go away. 22:16:12 You have to press spacebar to make it go away. 23:02:19 another in a long line of "monster dies before the attack is fully resolved" bugs 23:03:09 perhaps you could prevent these errors by changing the game so monsters can't be killed by attacks 23:44:25 Enclosed vault 13https://crawl.develz.org/mantis/view.php?id=12298 by Yermak