00:39:32 I'm having an absurdly weird bug I can't trace. I was restoring the old spell Twisted Resurrection in my fork and...everything spawned by it just disappears on reload. Monster version is unaffected, but player undead made this way just delete themselves. I can't find anything in tags that would clearly be causing this (my first thought was some 00:39:32 code from when the spell was removed auto-despawned the creatures, but...no?) I'm so confused right now. 00:58:47 Before and After reload (it seems so random in which ones it deletes): https://imgur.com/a/Y7IyVcO (I can't help but wonder if this happens to enemy ones too just more occasionally) 01:21:37 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-3303-gb091788534 01:27:15 Registering all of them as both God Gift (Yredelemnul) and as a Companion in the companions list finally fixed it...but...Yea I have no idea why they were being randomly deleted before; is there something in the code somewhere that doesn't like the player having summons of summon type 0 or something? It was truly bizarre behavior to me. 01:38:16 Unstable branch on crawl.develz.org updated to: 0.26-a0-113-gd8a95b691d (34) 02:16:33 Bcadren: if you have specific questions about mainline crawl behaviour I'm happy to answer them, but fork development is off topic 02:17:32 I suggest you use a debugger and add a breakpoint on monster removal to find out the cause 02:18:37 Straight up just reenabled an old spell that was disabled years ago and the monsters it spawned randomly got deleted on reload. Super weird. Whatever. 02:25:22 Ryzel (L10 TeWz) Crash caused by signal #6: Aborted (D:8) 02:27:55 that's not that surprising; in general crawl code is full of hidden assumptions and things tend to break over time 02:32:41 reloads are also not atomic: saving and loading are separate steps, and either could cause such an issue 02:36:00 Yea I found cases both where it wouldn't marshall a buggy monster and where it wouldn't unmarshall a buggy monster; but neither seemed to have affected it. I never figured out why it was happening, but registering the monsters as god gifts and god companions worked (which is fine since I'm restoring the old spell as a god power). 03:31:07 Fork (bcrawl) on crawl.kelbi.org updated to: 0.23-a0-3303-gb091788534 03:43:48 Fork (bcadrencrawl) on crawl.kelbi.org updated to: 0.22.1-2822-g80da662488 05:44:03 -!- amalloy is now known as amalloy_ 06:01:42 -!- amalloy_ is now known as amalloy 08:01:21 Just came here to report a bug on the tournament website (repo has no issue tracker): On the "All Players" leaderboard, if you click the arrows for sorting according to a category, you get sorted lexicographically by (string) rank, which is pretty useless data. 08:26:50 !kw nchoice 08:26:50 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 08:26:58 =nemelex HOAE 08:27:00 Permission kw:nchoice denied. 08:27:02 hmmm 08:27:18 this was not admin-only kw until fairly recently 08:32:02 (it still isn't) 08:33:46 yeah, I sent him an email 10:09:12 does anyone object if I reset taildb.log? it's 1.8G 10:55:17 advil: yeah, I saw how it has logs for past tournaments and I wasn't sure if this log would be useful to us in some way 10:55:31 if you think not, feel free 11:03:31 I think unlikely 11:10:44 Zagtar (L4 GrEE) Crash caused by signal #6: Aborted (D:3) 11:12:18 mti (L16 FoAK) Crash caused by signal #6: Aborted (Swamp:1) 11:12:28 freshwaterlynx (L3 VSIE) Crash caused by signal #6: Aborted (D:3) 11:12:48 Zagtar (L4 GrEE) Crash caused by signal #6: Aborted (D:3) 11:12:53 dweomer (L4 GrFi) Crash caused by signal #6: Aborted (D:3) 11:12:54 nickanc (L27 SpWz) Crash caused by signal #6: Aborted (Slime:2) 11:12:54 ggurman (L12 GrFi) Crash caused by signal #6: Aborted (Lair:4) 11:12:55 mti (L16 FoAK) Crash caused by signal #6: Aborted (Swamp:1) 11:13:05 Gearid (L5 OpBe) Crash caused by signal #6: Aborted (D:3) 11:13:06 CaptBlasto (L12 HaHu) Crash caused by signal #6: Aborted (D:11) 11:13:12 Graveyardigan (L12 TrEE) Crash caused by signal #6: Aborted (Lair:6) 11:13:22 advil: seems cao is having issues, are you working on it by chance? 11:14:26 GozertheGozerian (L4 GnSk) Crash caused by signal #6: Aborted (D:2) 11:14:27 Rangito (L15 DsWr) Crash caused by signal #6: Aborted (Depths:2) 11:14:33 KingOtar (L11 VSBe) Crash caused by signal #6: Aborted (D:9) 11:14:43 Gearid (L5 OpBe) Crash caused by signal #6: Aborted (D:3) 11:15:35 jtang582 (L7 HuCj) Crash caused by signal #6: Aborted (Sewer) 11:15:44 Zagtar (L4 GrEE) Crash caused by signal #6: Aborted (D:3) 11:15:58 jools (L21 GnWn) Crash caused by signal #6: Aborted (Elf:2) 11:16:47 Zagtar (L4 GrEE) Crash caused by signal #6: Aborted (D:3) 11:16:47 AntiSprigganMachine (L18 HOBe) Crash caused by signal #6: Aborted (Elf:1) 11:16:47 marcb3 (L18 MiFi) Crash caused by signal #6: Aborted (Lair:1) 11:16:56 Nuggy (L11 DrCj) Crash caused by signal #6: Aborted (D:10) 11:17:04 HANFGEIST (L27 GnCj) Crash caused by signal #6: Aborted (Abyss:4) 11:17:07 chrlz (L1 MiBe) Crash caused by signal #6: Aborted (D:1) 11:17:22 jtang582 (L7 HuCj) Crash caused by signal #6: Aborted (Sewer) 11:17:57 Prey (L15 FoAK) Crash caused by signal #6: Aborted (Orc:2) 11:17:59 marcb3 (L18 MiFi) Crash caused by signal #6: Aborted (Lair:1) 11:18:20 alphacentauri (L3 HOIE) Crash caused by signal #6: Aborted (D (Sprint)) 11:18:31 penpaper (L10 VSFi) Crash caused by signal #6: Aborted (D:8) 11:18:36 apparentbliss (L10 DDFi) Crash caused by signal #6: Aborted (D:9) 11:18:56 Daemonika (L16 DEFE) Crash caused by signal #6: Aborted (Shoals:1) 11:19:08 Prey (L15 FoAK) Crash caused by signal #6: Aborted (Orc:2) 11:19:10 AntiSprigganMachine (L18 HOBe) Crash caused by signal #6: Aborted (Elf:1) 11:19:20 jtang582 (L7 HuCj) Crash caused by signal #6: Aborted (Sewer) 11:19:35 sweetandcool (L10 DgCj) Crash caused by signal #6: Aborted (D:9) 11:19:36 penpaper (L10 VSFi) Crash caused by signal #6: Aborted (D:8) 11:19:56 chrlz (L1 MiBe) Crash caused by signal #6: Aborted (D:1) 11:20:34 mti (L16 FoAK) Crash caused by signal #6: Aborted (Swamp:1) 11:20:40 penpaper (L10 VSFi) Crash caused by signal #6: Aborted (D:8) 11:20:46 CaptBlasto (L11 HaHu) Crash caused by signal #6: Aborted (D:11) 11:21:26 mti (L16 FoAK) Crash caused by signal #6: Aborted (Swamp:1) 11:21:28 sweetandcool (L10 DgCj) Crash caused by signal #6: Aborted (D:9) 11:21:31 AntiSprigganMachine (L18 HOBe) Crash caused by signal #6: Aborted (Elf:1) 11:21:35 Nuggy (L11 DrCj) Crash caused by signal #6: Aborted (D:10) 11:25:52 uh 11:30:06 I'm not sure what happened but all of those games blocked on disk access at once 11:32:09 does anyone know who or what DissolvedMummy is? 11:33:15 never heard of it; someone scumming a mummy 11:35:31 well they stopped playing but the cpu usage was very high for some reason, and when I spectated for a sec it was behaving a bit like a bot in a tight loop 11:36:45 rc file doesn't look involved so it might have just been someone doing a lot of autoexploring while megazigging 11:39:09 my guess would be that cao just got overwhelmed 11:39:32 (impending disk failure for an outside chance) 11:40:18 noooo 11:40:48 clearly remove all loot (including dropped equipment) in zigs once they are sufficiently mega 11:50:30 yeah this might just be a player trying to scum zig ghosts for the unique category, they are already rank 6 despite having no completed games 11:50:49 combined with fast play in megazigs with a lot of items 12:44:03 Is Ozo Armour keeping the buff when you're being trampled bug or feature? 12:45:32 feature i think 12:46:11 it's intentional that it doesn't break on other ways of moving around like blink or hop or whatever since those wouldn't be particularly abusable, and it feels neater to just have it be explicitly player movement that breaks it rather than also add a long list of other special-cased types of movement that may or may not break it 13:11:20 thanks 13:36:16 I'm not actually sure that anyone here but me would have any ideas, but...does anyone have ideas about why on cao fast(?) tabbing is sometimes leading to `error: [Errno 11] Resource temporarily unavailable`? 13:36:24 I think this is new 13:38:50 any stack trace for that? 13:40:21 I'm not sure there's much to be learned from it, but here's a sample: https://pastebin.com/CXMB8smB 13:40:38 one theory I've had is simply that the tabs are coming so fast that the socket is still blocking 13:41:02 given that cao is under extremely heavy load 13:41:40 haha actually those errors in that log appear to be just a few ms apart, though that's not typical 13:42:13 hm the player I was just looking at now has a pretty involved rc file too, http://crawl.akrasiac.org/rcfiles/crawl-0.25/Boheme.rc 13:47:51 those messages are 25 bytes, and the socket has a send buffer of 2048 bytes, so it'd take around 80 tabs for it to fill 13:48:55 i'd imagine that the small time gap is because they got bunched together before being processed 13:53:36 yes, it's usually more like 80-100ms 13:55:04 hm, it doesn't seem entirely impossible that someone holding down tab with a very high key repeat in some scenarios could hit that limit 13:58:45 seems plausible 13:59:08 is the load high enough that there's visible lag? 13:59:09 aha 13:59:24 yeah I replicated it by shooting arrows at a test statue holding down tab 14:00:28 I think people have been experiencing higher lag than normal, but I'm not sure it's worse than typical for t 14:00:46 I've also definitely been seeing recent reports that firing things is laggy 14:01:30 ok, so if this is what is doing it then dropping some of the tabs does seem like perfectly reasonable behavior 14:01:43 (just holding tab in melee wasn't doing it) 14:02:36 this seems like a thing that webtiles should be doing 14:02:47 although i guess there's also dgamelaunch, urgh 14:03:03 tab at least is probably a relatively safe key to drop 14:03:38 yes, what I mean is that anyone who is seriously playing in the way that I just simulated is not going to notice if some tabs get lost while the socket blocks 14:04:26 right 14:04:35 I'm not sure I think webtiles needs to do any more? It might conceivably be nice on the python side to notice this and throttle or pause key input for some time maybe 14:04:42 since this is also relatively high cpu 14:05:30 i was thinking more that dropping any old key could cause players to make unintended moves, e.g. if an initial menu-opening key was dropped 14:06:16 i'd rather have it in webtiles than add more ifdef gunk 14:13:12 ok DissolvedMummy is just someone playing fast in zigs who has hit a bug I think, they were very responsive and will take a break 14:13:16 it may be a gold bug 14:14:37 -!- amalloy is now known as amalloy_ 14:17:59 what I had in mind is if the socket has an errno 11 (which I'm pretty sure in a tornado context only means that the socket will block) then python should ignore input from webtiles for 500ms or something 14:19:58 that socket has nothing to do with tornado, afaik; it's just a regular socket 14:20:37 but yes, that is what it would mean, and that does sound reasonable 14:22:19 on the other hand, just dropping the key is probably enough; there's no need for a 500ms timeout since the socket will regulate its own send buffer (via eagain) 14:22:54 the throttling would be for cpu sake 14:23:28 that is, when this happens it corresponds with a cpu spike on the part of the crawl binary 14:23:55 it's already effectively dropping the key I believe 14:24:34 it seems like a strange way to throttle cpu 14:24:57 although i do realize that it's not easy to deploy more effective and straightforward ways :/ 14:25:22 it actually seems better than many of the more direct ways I've come up with! 14:25:30 though it only covers rare cases 14:25:41 i was thinking cgroups in this instance 14:26:41 although there is a certain elegance to it.. if crawl isn't keeping up, the queue will naturally fill 14:27:47 I should look into cgroups, I think last time this was discussed I didn't have access to try out something like that 14:28:03 would also help with stuff like the megazig case that just came up 14:40:54 cgroups still let processes burn 100% cpu if there's no other processes using it, so it's not perfect 17:58:14 does anyone know why CBL has a hard constraint requiring demonic holiness in _ms_waste_of_time? 17:58:32 having trouble figuring out why it would be implemented this way 18:00:01 could be some weird way of making it so that pan lords can cast it but player ghosts can't? 18:00:23 that's about all that I came up with too 18:00:41 can you poly lom lobon? it would keep poly'd lom lobon from casting it 18:01:18 you can't poly &s 18:01:35 shows you how much I ever have used poly 18:01:43 (although if you could they'd stay demonic anyway, so could still cast it right?) 18:02:03 wouldn't they inherit the holiness of the new form? 18:02:33 polymorph is supposed to always keep the same holiness, so demons poly into other demons of the same tier 18:03:09 ah, did not know that either 18:04:11 it used to only work on natural monsters but then at some point in the distant past got extended to demons (and holies? i don't remember) 18:04:46 but yeah, it only ever really comes up for demons in cases like xom acting 18:19:57 and grinder the hellion 22:20:15 Er, if I read it right, it means that a demonic being will never _not_ cast it; it cares not if you are friendly to it and vulnerable, hence the comment. 22:33:00 it's && so it will always be a waste of time for non-demons 22:33:06 that comment is pasted from elsewhere I think 22:33:33 No, the function checks if it _is_ a waste of time. 22:33:46 yes, it's a waste of time if ! demonic 22:33:47 So because demons are rude, they'll always let it off. 22:34:18 Hang on, one of us is confused and it might be me 22:34:44 It's a waste of time if the monster is friendly _and_ you are vulnerable or low on HP _and_ the monster is not demonic. 22:35:05 So, yes, what I said - demons are rude and will always let it off, no matter your HP. 22:35:57 An !demonic monster will not cast it because it likes you and it might hurt you badly. 22:38:13 ah right 22:40:36 now I don't know why it's a waste of time if you are relec 0 22:40:59 oh 22:41:15 no, I just keep getting perpetually confused by the inverted logic 22:43:08 Yeah, me too, I had to take several looks at it before I decided it was right. 22:43:32 Also 090cc2b85fdcd05118047d191841931eee2903af has the most useful commit message ever. 22:43:48 having an exception here for demons strikes me as kind of silly, but w/e 22:46:40 -!- amalloy_ is now known as amalloy 22:48:02 Whether the design is right is up to you lot, but I think it does implement the design.