00:01:29 but maybe gammafunk knows something I don't :) 00:02:36 I'm assuming objstat and so forth runtime are domianted by the PRNG code 00:02:43 maybe I'll try to validate that later 00:18:31 ranchugoldfish (L15 DsFi) ERROR: range check error (-1 / 53) (D:14) 00:30:27 03DracoOmega02 07* 0.33-a0-713-g752c899f1e: Fix body armour being tagged {auxiliary armour} for stash search 10(2 hours ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/752c899f1ebe 00:30:27 03DracoOmega02 07* 0.33-a0-714-g535689ca4f: Make Freeze use a proper zap 10(44 minutes ago, 9 files, 21+ 22-) 13https://github.com/crawl/crawl/commit/535689ca4f4d 00:30:27 03DracoOmega02 07* 0.33-a0-715-g233abfc1fd: Fix one's armour sometimes silently falling off (pisaster) 10(2 minutes ago, 1 file, 2+ 2-) 13https://github.com/crawl/crawl/commit/233abfc1fd47 00:33:43 Unstable branch on cbro.berotato.org updated to: 0.33-a0-712-gc0115c06c0 (34) 00:44:32 03DracoOmega02 07* 0.33-a0-716-gdc257c63a9: Replace an assert with a prompt instead 10(2 minutes ago, 1 file, 2+ 1-) 13https://github.com/crawl/crawl/commit/dc257c63a9f3 00:44:32 03DracoOmega02 07* 0.33-a0-717-ga3f7f667d0: Remove a breadswinging-related comment 10(68 seconds ago, 1 file, 0+ 4-) 13https://github.com/crawl/crawl/commit/a3f7f667d0aa 00:55:28 Monster database of master branch on crawl.develz.org updated to: 0.33-a0-715-g233abfc1fd 04:32:27 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-5249-g4a8afe7061 05:32:00 Unstable branch on crawl.akrasiac.org updated to: 0.33-a0-717-ga3f7f66 (34) 08:01:16 <08n​icolae> i have this problem irl 08:01:40 <09g​ammafunk> re objstat computation, I'm not sure what it would mean for it to be dominated by rng. The actual calls to rng would be a very tiny fraction of execution time. It's running through the entirety of dungeon generation, creating layout, placing vaults, generating monsters and items etc. Since it's doing the exact same thing as the crawl binary that's done during a normal game, that's just lots of time spent on the general logistics of 08:01:40 creating items 08:02:24 <09g​ammafunk> we don't have any abstraction that can compute only "values" of objects without creating the actual objects first 08:04:18 <09g​ammafunk> but in general regarding running objstat, I think it's probably not worth setting up something in terms of a dedicated server just to run objstat. There's value in having maybe weekly objstat runs of trunk, and I do have plans to do this once I move my bots from their current amazon lightsail instance to a lower tier (cheaper) digital ocean droplet 08:04:56 <09g​ammafunk> but I have to update some code for changes to newer python libraries first 08:05:32 <09g​ammafunk> my vague plan was just to run say 500 iterations of trunk objstat on said digital ocean instance on a weekly basis 09:30:31 <05i​coson> I'm guessing what they mean is the levelgen aspects that are rng driven 09:56:34 Should I make objstat a github CI action? 09:57:12 <09g​ammafunk> I'm not sure if having it run some smallish number of iterations would be useful 09:57:30 <09g​ammafunk> perhaps running objstat and mapstat both a small number of iterations could catch some bugs 09:57:50 <09g​ammafunk> objstat (and probably mapstat as well) get broken from time to time 09:58:04 <09g​ammafunk> unfortunately sometimes the bugs take a large number of iterations to see, like 09:58:20 <09g​ammafunk> %git f801aa8b2f4d7fc6510385ecc00aee12d295ffde 09:58:21 <04C​erebot> gammafunk * 0.33-a0-304-gf801aa8b2f: Fix mapstat/objstat initialization issues (9 weeks ago, 18 files, 39+ 42-) https://github.com/crawl/crawl/commit/f801aa8b2f4d 09:58:39 <09g​ammafunk> was only caught when I tried to run the official objstat results of 1k iterations 09:59:06 How long does 1k iterations take? 09:59:12 <09g​ammafunk> a long time 09:59:45 Hmm. 09:59:45 <09g​ammafunk> on a 2 vcpu lightsail instance, the entire official data set took: > real 2253m54.826s > user 4145m9.874s > sys 7m25.299s 09:59:57 <09g​ammafunk> but that's 4 different 1k datasets 10:00:22 <09g​ammafunk> however if you're looking for comprehensive data, you need those 4 datasets 10:00:26 I think the time limit of github actions is like 6 hours? 10:00:40 <09g​ammafunk> I thought it was much shorter than that even 10:00:55 <09g​ammafunk> in the past it was like 10 minutes or something, maybe they've change that 10:01:01 My actions in another project routinely hit 90-120 minutes 10:01:07 and they are somehow fine 10:01:15 never saw timeouts. 10:01:20 <09g​ammafunk> I'm not sure if a weekly objstat of trunk with a much smaller number of iterations could work in github actions 10:01:42 I think 6 hours is probably not good enough 10:01:44 <09g​ammafunk> if they give you e.g. 4 cores or more then it's possible and could be very nice to have that automated potentially 10:01:52 <09g​ammafunk> yeah like 10:02:12 <09g​ammafunk> !calc 2254 / 2.0 / 60 10:02:14 <04C​erebot> 18.78 10:02:33 <09g​ammafunk> if you had 4 vcpu on github you'd probably need 20-ish hours 10:02:44 <09g​ammafunk> for 1000k 10:03:00 <09g​ammafunk> or not 1000k, but 1k 10:03:07 <09g​ammafunk> so if you divide that by 5, down to 200 iterations maybe it can work 10:03:36 <09g​ammafunk> lightsail vcpus get throttled a lot with high utilisation and I'm not sure how github's system would be over a long period 10:04:12 <09g​ammafunk> but yeah maybe there's value in a CI action for just running objstat and mapstat on 10 iterations for a single "all levels" dataset 10:04:28 <09g​ammafunk> in theory it could catch a bug or two 10:08:49 <09g​ammafunk> Anyone wanting to mess with objstat, CI or otherwise, might find https://github.com/gammafunk/objstat useful, specifically https://github.com/gammafunk/objstat/blob/master/objstat/objstat.sh and https://github.com/gammafunk/objstat/blob/master/objstat/objstat-job.sh 10:11:16 <09g​ammafunk> also there's the complication of getting the TSV files into a more useful format like google sheets, which I have https://github.com/gammafunk/objstat/tree/master/objstat/UploadObjstat for that, which is effective but the interface is extremely crude 10:11:33 <09g​ammafunk> and needs to be reworked to support some kind of better google api for submitting the tsv files 11:25:44 New branch created: dependabot/pip/crawl-ref/source/webserver/requirements/virtualenv-20.26.6 (1 commit) 13https://github.com/crawl/crawl/tree/dependabot/pip/crawl-ref/source/webserver/requirements/virtualenv-20.26.6 11:25:45 03dependabot[bot]02 {GitHub} 07[dependabot/pip/crawl-ref/source/webserver/requirements/virtualenv-20.26.6] * 0.33-a0-718-g652b7ab1f4: Bump virtualenv in /crawl-ref/source/webserver/requirements 10(65 seconds ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/652b7ab1f49f 11:25:46 New branch created: pull/4236 (1 commit) 13https://github.com/crawl/crawl/pull/4236 11:25:46 Branch pull/4236 updated to be equal with dependabot/pip/crawl-ref/source/webserver/requirements/virtualenv-20.26.6: 13https://github.com/crawl/crawl/pull/4236 13:00:21 BerryKnight (L2 DgCj) ASSERT(item_type_is_equipment(item.base_type)) in 'player-equip.cc' at line 407 failed. (Ossuary) 13:00:51 BerryKnight (L2 DgCj) ASSERT(item_type_is_equipment(item.base_type)) in 'player-equip.cc' at line 407 failed. (Ossuary) 13:01:46 BerryKnight (L2 DgCj) ASSERT(item_type_is_equipment(item.base_type)) in 'player-equip.cc' at line 415 failed. (Ossuary) 13:17:09 <03i​mplojin> !crashlog berryknight 13:17:14 <04C​erebot> 8. BerryKnight, XL2 DgCj, T:516 (milestone): https://cbro.berotato.org/morgue/BerryKnight/crash-BerryKnight-20250113-200139.txt 13:17:15 <03i​mplojin> is this more inscription shenanigans 13:18:23 <03i​mplojin> oh the server just hasn't updated yet 13:18:25 <04d​racoomega> It's the bug I fixed last night 13:18:27 <04d​racoomega> Same person, even 13:18:45 <04d​racoomega> I don't know if they are doing this deliberately or what 13:18:54 <03i​mplojin> yeah i was just wondering if it was a different form of it 13:20:38 <03i​mplojin> berryknight is a longtime player, you'd think they'd realize what was happening, but who knows 13:24:35 <04d​racoomega> (I can't help but wonder if someone in an Ossuary at xl 2 is actually doing this to reset things ^^; ) 13:25:55 <05i​coson> I think sometimes even very experienced players will just keep trying a bug repeatedly until it goes away, they don't know when servers update etc (and don't realize we get a log of it) 14:26:19 03Implojin02 07* 0.33-a0-718-g836663c12a: Add Charlie Gettys to the CREDITS 10(87 seconds ago, 1 file, 1+ 0-) 13https://github.com/crawl/crawl/commit/836663c12a83 16:41:21 Unstable branch on underhound.eu updated to: 0.33-a0-718-g836663c12a (34) 16:42:51 BerryKnight (L2 DgCj) ASSERT(item_type_is_equipment(item.base_type)) in 'player-equip.cc' at line 415 failed. (Ossuary) 17:53:02 03DracoOmega02 07* 0.33-a0-719-g12a0c8464c: Actually stop the Sword of Zonguldrok from generating 10(6 hours ago, 1 file, 1+ 0-) 13https://github.com/crawl/crawl/commit/12a0c8464c4c 17:53:02 03DracoOmega02 07* 0.33-a0-720-g6f71aec649: Don't warn about fleeing apostle challenges for entering a PoG (salticid) 10(6 hours ago, 1 file, 4+ 1-) 13https://github.com/crawl/crawl/commit/6f71aec6494b 17:53:02 03DracoOmega02 07* 0.33-a0-721-gdfcb6f04d8: Fix being unable to unwield a magical staff (ragingrage) 10(3 hours ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/dfcb6f04d8d8 17:53:02 03DracoOmega02 07* 0.33-a0-722-g812cd53223: Fix enhancer staves not getting the usual wield message 10(3 hours ago, 1 file, 2+ 7-) 13https://github.com/crawl/crawl/commit/812cd5322372 17:53:02 03DracoOmega02 07* 0.33-a0-723-g60363ce734: Fix religious abilities having the wrong hotkeys for zealot starts (Kab) 10(2 hours ago, 2 files, 4+ 9-) 13https://github.com/crawl/crawl/commit/60363ce73448 17:53:02 03DracoOmega02 07* 0.33-a0-724-g9e0d956b5c: Make a few more item types count as always identified 10(89 minutes ago, 1 file, 3+ 0-) 13https://github.com/crawl/crawl/commit/9e0d956b5c5d 17:53:02 03DracoOmega02 07* 0.33-a0-725-g77dd6fc368: Prevent a crash when Surprising Crocodile drags a monster into a shaft 10(88 minutes ago, 1 file, 9+ 5-) 13https://github.com/crawl/crawl/commit/77dd6fc368cb 17:53:02 03DracoOmega02 07* 0.33-a0-726-g203f4b1603: Don't let the player kill themselves with Makhleb marks + vitrify 10(48 minutes ago, 3 files, 6+ 4-) 13https://github.com/crawl/crawl/commit/203f4b1603e4 20:08:44 CI action was what I was proposing. Though I see Gammafunk's point about iteration count 20:10:28 Another interesting thing to do there would be to run n iteraitons on different targets 20:10:53 if you pick the dungeon seeds deterministically (say by checking in a file once that's always used) 20:11:06 then all the targets should produce exactly same stats 20:11:10 or we have a levelgen bug 20:11:23 e.g. sequence points 20:11:31 Ik you guys have some other tests that attack that too a bit but you get my point 20:13:14 As for this @icoson: Ik you guys have some other tests thatI think sometimes even very experienced players will just keep trying a bug repeatedly until it goes away, they don't know when servers update etc (and don't realize we get a log of it) 20:13:32 maybe they do know 20:13:42 and want to make sure devs notice :P 20:16:47 <04d​racoomega> Well, the thing about that crash I was referring to earlier today is that even if it didn't crash, it couldn't possibly do anything 20:17:09 <04d​racoomega> I mean, maybe they were deliberately trying to wield the scroll in a weird way, for funsies? 20:17:16 <04d​racoomega> But it seems strange to pause the game in place until being able to do so 20:18:18 cat on keyboard? 20:19:34 <04d​racoomega> (Also, you say "If we pick the seeds deterministically, all targets should produce the exact same stats", but the game isn't seed stable between versions. All sorts of changes can cause a seed in one minor version to produce a completely different dungeon in another version - edits to vaults, additional or removal of items, item weights, etc.) 20:20:03 <04d​racoomega> So it is in fact expected that the same seed will not keep reproducing the same result as new things are committed 20:20:14 <04d​racoomega> (It just has to produce the same result on the same version) 20:20:14 I wasn't saying version over version 20:20:33 Right, I'm saying you could use CI to validate that property on every commit 20:20:39 at least to some extent 20:21:17 Would require somewhat involved github actions setup, granted 20:21:26 but it's feasible to do 20:21:33 <04d​racoomega> I mean, it can change from commit to commit, even 20:21:38 Yeah, I know 20:22:01 I was saying, you could check clang/gcc/ whatever version of clang mac is using / so on 20:22:07 for the same commit 20:22:21 just like you build and header test them all 20:23:25 <09g​ammafunk> I think we have a lua test the we do run already for seed stability 20:23:36 <09g​ammafunk> as part of the make test suite 20:23:40 <09g​ammafunk> it's a custom lua thing that advil wrote 20:24:09 <09g​ammafunk> it doesn't try to handle different targets, but that's probably more because that would rarely be a source of bugs 20:27:13 I think I came across reference to that 20:27:30 there's also a vaguely new gcc warning of dubious helpfulness 20:27:34 -Wsequence-points 20:29:07 I need to put together a PR conditionally turning it on 20:29:59 anyway, if you do it on every commit, even if you gather less data per commit, you'll end up with trend data 20:30:30 if one is sneaky with data analysis 20:30:35 <06w​ensley> to be safe, update tag_major_version every time you change anything that might affect seed stability 20:31:18 :D 20:32:27 just update it every commit for giggles 20:32:27 append-only codebase 20:34:12 Anyway, on another topic 20:34:46 I'm happy to do various "boring" sounding things 20:34:46 like removing all unnecessary includes 20:35:03 which might help builds, especially incremental ones 20:35:14 which might give dracoomega's potato a break :) 20:38:51 Or I could attack how much spooky action at a distance there is / try to make the startup sequence unit testable 20:38:51 But while I do enjoy doing these sorts of things, the joy for me is seeing the improvemnt actually land. Ik, everyone's busy, not asking for promises that it'll be merged any tiem soon 20:38:51 But I'm hesitant to embark on such quests unless folks want to see those things happen / will maybe have time to reviewthem and merge them 20:39:56 03DracoOmega02 07* 0.33-a0-727-ge79d8a169f: Fix a crash with the known items menu 10(2 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/e79d8a169f3f 20:45:01 let me see if I can remove all the necesseary headers and y'all can tell me what you think, that one should be easy enough 20:57:15 03Cgettys02 07https://github.com/crawl/crawl/pull/4233 * 0.33-a0-714-g2ad3caa5dd: Unused header cleanup 10(3 minutes ago, 3 files, 0+ 22-) 13https://github.com/crawl/crawl/commit/2ad3caa5dda2 20:57:28 Arg, forgot to change branches, sorry for the spam 20:59:27 03Implojin02 07* 0.33-a0-728-gaf1eef3400: Always identify a couple more base item types 10(2 minutes ago, 1 file, 4+ 2-) 13https://github.com/crawl/crawl/commit/af1eef3400e5 21:25:24 03hellmonk02 07[shoppricing] * 0.33-a0-651-ga9f2c6ea03: Make Gozag shop prices independent of xl 10(2 minutes ago, 1 file, 5+ 2-) 13https://github.com/crawl/crawl/commit/a9f2c6ea0376 21:40:02 quagga (L2 FoGl) Crash caused by signal #6: Aborted (D:1) 21:48:58 Or maybe not, given all the ifdefs involved :D 22:00:01 03Implojin02 07* 0.33-a0-729-g535f1cdcc7: Don't prevent item stacking based on differing inscriptions 10(31 minutes ago, 1 file, 0+ 8-) 13https://github.com/crawl/crawl/commit/535f1cdcc7a6 23:18:27 <03i​mplojin> i'm going to merge that remove glu dependency PR, it seems fine to me 23:20:00 03Cgettys02 {Implojin} 07* 0.33-a0-730-gde7caedde6: Remove GLU dependency 10(4 days ago, 8 files, 100+ 17-) 13https://github.com/crawl/crawl/commit/de7caedde691 23:35:27 Unstable branch on crawl.develz.org updated to: 0.33-a0-730-gde7caedde6 (34) 23:47:54 I must have missed your last message by like 30 seconds :D 23:48:12 now we just see if I somehow broke som exotic target despite every precaution 23:48:22 prett sure it'd require a non-spec compliant OpenGL implementation 23:48:26 but you never known 23:51:02 <03i​mplojin> yeah, i looked at it the other day after that back and forth and i'm pretty sure it's fine 23:51:04 <03i​mplojin> /users 23:51:14 <03i​mplojin> if it's not for some reason, we can fix it 23:58:32 Windows builds of master branch on crawl.develz.org updated to: 0.33-a0-730-gde7caedde6