00:44 -!- raisse has joined #hardfought 00:44 -!- mode/#hardfought [+v raisse] by ChanServ 01:21 -!- elenmirie has quit [Ping timeout: 240 seconds] 01:27 -!- elenmirie has joined #hardfought 01:27 -!- mode/#hardfought [+v elenmirie] by ChanServ 01:38 -!- LarienTelrunya has joined #hardfought 01:41 -!- LarienTelrunya_ has joined #hardfought 01:45 -!- LarienTelrunya has quit [Ping timeout: 260 seconds] 01:45 -!- LarienTelrunya_ is now known as LarienTelrunya 01:51 !tell K2 another bugfix for slex, now complete with changelog :) can you update? thanks! 01:51 Will do, LarienTelrunya! 01:54 -!- jonadab has quit [Ping timeout: 240 seconds] 01:56 -!- Crawldragon has joined #hardfought 02:01 -!- jonadab has joined #hardfought 02:01 -!- mode/#hardfought [+v jonadab] by ChanServ 02:05 [dnh] starboy (Ana Dwa Mal Cha), 27832 points, T:6893, killed by an owlbear 02:45 -!- NeroOneTrueKing_ has joined #hardfought 02:49 -!- NeroOneTrueKing has quit [Ping timeout: 248 seconds] 03:20 -!- morgoth__ has joined #hardfought 03:23 -!- elenmirie_ has joined #hardfought 03:23 -!- elenmirie has quit [Read error: Connection reset by peer] 03:23 -!- elenmirie_ is now known as elenmirie 03:58 -!- noty has quit [Quit: WeeChat 1.9] 04:02 -!- Crawldragon has quit [Ping timeout: 246 seconds] 04:58 -!- morgoth__ has left #hardfought 05:25 -!- mode/#hardfought [+v elenmirie] by ChanServ 06:24 YANI/YASI: overhaul the algorithm used for determining which message will be on a t-shirt. I'd want to make it so that it's randomized for every t-shirt, instead of depending on the o_id which would give predictable messages to two t-shirts generated shortly after each other. But how? 06:56 Slash'EM Extended updated 06:56 K2: Message from FIQ at 2017-09-05 23:13 EDT: pushed an update that fades overprotection rapidly 06:56 K2: Message from LarienTelrunya at 2017-09-06 01:51 EDT: another bugfix for slex, now complete with changelog :) can you update? thanks! 06:57 FIQhack updated 06:57 K2: yay, many thanks! :) 07:05 o/ 07:11 LarienTelrunya: Use the MD5 hash of the o_id 07:12 Or maybe just game_id / o_id 07:12 Or game_id + o_id for that matter 07:13 isn't game_id constant for a given character? 07:13 No, wait, game_id + o_id solves a different problem. Not the one you were asking about. So not that. 07:13 Yes, it's constant within a given game. 07:13 or maybe not constant upon saving and loading 07:13 Eh, not sure about non-NH4 07:13 In NH4 it's constant. 07:14 maybe I should just add a new int field to obj.h, which every item gets, and which is a rnd(1000000) upon generation of an item, which is then used for determining the shirt message 07:14 Seems unnecessary. 07:14 game_id / o_id should be fine. 07:15 It should be constant for a given object, but will not be adjacent numbers for sequentially generated objects, necessarily. 07:15 Technically it's predictable, but only if you know the game_id and the o_id, so actually not really predictable for normal players. 07:16 Not that anyone who plays slex could really be called a "normal" player, but you know what I mean. 07:18 K2: uh-oh... the hardfought server kind of "went out" for me :O am I the only one? can you still access it? 07:19 I've got a FIQHack game loaded on hdf now. 07:19 jonadab: hmmmmmm... maybe using something else than game_id might be the better choice 07:19 ah now the server is working for me again too 07:19 K2: false alarm, sorry :) 07:22 hah 07:23 gotta learn how to do a traceroute 07:23 quick/fast way to see if there's a network issue between you and whatever it is you're trying to access 07:24 I think windows calls it tracert 07:24 wow, didn't know such a thing existed :) it works! I'll use it from now on if I can't connect 07:27 ping and traceroute are the two most fundamental network diagnostic tools. 07:30 ping I already used to check how much lag I have 07:30 didn't know about tracert though 08:09 !lastgame dolores 08:09 LarienTelrunya: https://www.hardfought.org/userdata/d/dolores/slex/dumplog/1504643041.slex.txt 08:16 [nd] APic (Wiz Elf Mal Cha) polymorphed his first object, on T:41477 08:22 fiq: Error: vision calculations during level creation: an invisible archeologist->an aligned priest Location: libnethack/src/prop.c:2361 08:22 P does not repair this 08:23 R does removes the error but the next move causes it again 08:25 i had to use Q. game must be recovered manually 08:25 k2: please note 08:27 i have to leave. hope this isn't too hard to fix 08:27 -!- LarienTelrunya has quit [Quit: Page closed] 08:42 ok 08:42 FIQ ^^ 08:46 😸 08:54 hothraxxa: "game must be recovered manually" may actually just mean "wait for FIQ to fix the bug". 09:39 [gh] Tangles (Val Dwa Fem Law) wished for "uncursed magic marker", on T:18014 09:40 !whereis Tangles 09:40 K2: Tangles [gh]: (Val Dwa Fem Law) T:17931 The Dungeons of Doom level: 20 09:58 EPI: any randomly generated object can randomly ungenerate (just cease to exist) at any time. 09:58 The RNG giveth and the RNG taketh away. 10:00 addendum: this is guaranteed to happen within a certain number of turns after generation 10:03 Eh, I think guarantees are less evil than probabilities. 10:17 EPI: whenever a player gains a point of Luck, another random player on the server loses a point of Luck without being notified. If no other players are on the server, the player who gained the point immediately loses it. 10:18 [gh] Tangles (Val Dwa Fem Law) wished for "blessed greased fireproof +5 dragonhide t-shirt of displacement named Gnomish Mines Drowning Survivor", on T:18072 10:19 You can't read the shirt while wearing it. 10:19 That's dumb 10:20 It reads: "Furinkan High School Athletic Dept." 10:20 Hmm... Shoulda called it Gnomish Mines Swim Team or someting. 10:22 Heh. 10:24 [gh] Tangles (Val Dwa Fem Law) wished for "blessed fireproof +5 oilskin dragonhide fedora named Indy <3", on T:18096 10:26 That's 0 from 4 on the +5, but my AC is below -50 anyway. 10:37 [gh] Tangles (Val Dwa Fem Law) wished for "blessed fireproof +5 dragonhide gauntlets of dexterity and power named Like... a GLOVE!", on T:18103 10:53 [gh] Tangles (Val Dwa Fem Law) had Excalibur thrown at her by some watery tart, on T:18691 10:56 [gh] Tangles (Val Dwa Fem Law) killed Medusa, on T:18820 11:03 hrm. i guess this means my game is still borked 11:03 hi 11:03 Hello FIQ, Welcome to #hardfought 11:03 oh look a fiq 11:03 i have a busted game 11:03 yes 11:04 i think it can't generate the sanctum properly 11:04 are you getting a "this game needs manual recovery" error? 11:04 no. it's the standard three choices, but only Q works 11:04 ah 11:04 then K2 doesn't need to help 11:04 but yeah I'll need to fix it either way 11:04 P does nothing. R removes the error but trying to go down causes it to happen again 11:04 what happens if you press P? 11:05 uh 11:05 P probably does something but the error happens again 11:05 What happens if you press P repeatedly? 11:05 right 11:05 and will probably work eventually 11:05 but there's a bug either way 11:05 want me to try repeated P? 11:05 I think what happens is that the priest has some property that causes a vision calculation 11:06 agggravate monster, invisibility, or whatever 11:06 Wasn't the Arc invisible, and don't aligned priests have see invisible? 11:06 I tried to remove all cases of sense checks during vision recalc from this but must have missed one 11:06 jonadab: invisible arc is presumably hothraxxa 11:06 it is me 11:06 the game is checking if hothraxxa can see an aligned priest 11:07 during level gen 11:07 Ah. 11:07 Can a priest be generated with an item that grants displacement or something, due to properties? 11:07 yes 11:07 this is likely t o be what is going on 11:07 i have see invisible! i can cast astral sight! i can see anything! 11:07 I have encountered similar issues before 11:07 I tried to fix all cases of this happening 11:07 but I probably forgot one case 11:10 i should get a fiqhack badge for bug detecting 11:13 Fixed the bug 11:14 yay 11:14 hothraxxa: However 11:14 You should be able to press P repeatedly to dismiss it 11:14 Before waiting for K2 11:14 K2: ^ 11:14 ok 11:14 ? 11:14 let's see what happens 11:14 K2: I fixed the bug 11:14 to push the fix 11:14 ok 11:14 You can update fh 11:14 ok 11:15 i'll wait for that then 11:15 hothraxxa: yeah I wasn't aware that K2 was on 11:15 that's why I said you could just spam P to get past it :P 11:15 Sorry 11:15 why are you apologizing? 11:15 you're secretly canadian aren't you 11:16 lol 11:16 FIQhack updated 11:16 tyhanks guys 11:16 ok 11:16 there's an amulet calling me 11:16 also, fh loads games amazingly fast now 11:17 hothraxxa: There may or may not be a drawback 11:17 Saves will also likely be bigger now 11:17 THe question is "how much bigger" 11:17 *The 11:17 I made it do binary checkpoints 10 times more often 11:18 To cut down on loading times and replaymode/etc 11:18 running errand be back in an hour or so 11:18 k2 might care. doesn't impact me 11:18 well it does. it makes game-loading nice 11:18 but i have no space issues ;) 11:19 heh 11:19 will have to see 11:19 yeah Iwas going to check once your game is done 11:19 how much impact will it have on cpu use also 11:19 (saves are made public when ready) 11:19 K2: uh, CPU time will benefit greatly 11:19 the major drawback is space 11:19 how much space does it take 11:19 oh i see 11:19 gotcha 11:20 saves are no big deal, those go away once you die/asc/quit 11:20 please don't delete saves if you don 11:20 't have to 11:20 is this going to increase ttyrec file size? 11:20 not at all 11:21 it will technically decrease it 11:21 since there's no loading screen if saves load faster 11:21 neat 11:21 but the difference is marginal 11:21 and yeah there's no reason for me server side to delete saves 11:21 K2: well what I am wondering is 11:21 unless requested due to a bug or issue etc 11:21 if saves are taking up like 200MB instead of 7 11:22 thats fine 11:22 then I should see if I can change up the chechkpointing a bit 11:22 i have GB of free space on the server SSD still 11:22 but if it's 20MB instead of 7 11:22 GB's 11:22 then it is *probably* worth it 11:23 either way this is probably a stopgap fix 11:23 so starting a brand new fh game 11:23 and saving on turn 1 11:23 until I can fix perf by making it try the nitrohack approach to save loading 11:23 15k 11:23 K2: that should be unchanged 11:23 basically 11:24 every n turns or so 11:24 :) 11:24 the game does a full save instead of a diff 11:24 ah 11:24 how many turns would be a good test 11:24 the "n" here 11:24 was changed 11:24 to be 10 times more often 11:24 so that number is... ? 11:24 there is no specific turncount, it uses size calculation to figure out when to make a diff 11:24 and this is why I'm not sure on the actual effect 11:24 on save size 11:24 i see 11:25 so we need a game that has a fair amount of progression 11:25 fiq: i still get displacement removal errors 11:25 ok i *have* to step out now, i will be back soon 11:25 this time it was a zombie reviving 11:26 hm 11:26 so any case of revival does weird things to displacement 11:26 got it 11:29 yes, it just happened with a vampire as well 11:36 The invisible renegade priestess of Moloch reads a scroll of genocide! 11:36 The invisible renegade priestess of Moloch has found a scroll of genocide! 11:36 wrong order 11:38 =] 11:47 -!- LarienTelrunya has joined #hardfought 12:11 -!- raisse has quit [Ping timeout: 246 seconds] 12:13 hrm. spell maintenance and the amulet don't mix well 12:18 the amulet doubles maint c ost 12:18 *maint cost 12:30 -!- noty has joined #hardfought 13:02 [fh] hothraxxa (Arc Hum Fem Neu) wished for "uncursed historic greased magic marker named damon runyon", on T:49688 13:33 -!- raisse has joined #hardfought 13:33 -!- mode/#hardfought [+v raisse] by ChanServ 13:40 -!- Crawldragon has joined #hardfought 14:26 [fh] hothraxxa (Arc Hum Fem Neu) entered the Planes, on T:50172 14:27 silly bugger cast double trouble immediately 14:28 how the bouncing death ray missed tangles i'll never know 14:32 Bugbears pwn. 14:33 you hit the silly bugger. The silly bugger hits. You die. 14:33 go hothraxxa! 14:42 i had to abandon sirius fudd 14:48 [fh] hothraxxa (Arc Hum Fem Neu) reached the Astral Plane, on T:50349 14:57 curse you fiq 14:59 ? 14:59 ! 14:59 wand rebalance 15:00 ! 15:00 ^^ 15:00 9😸 15:05 l - a blessed +50 ring of protection (reflect,search,hunger,stealth,telepathy,warn,aggravates,dex,brill) (on left hand). 15:06 lol 15:07 lol 15:18 hothraxxa: wand rebalance is awesome on astral. If you have expert, you can light the whole plane with a single zap of blessed wand of light. 15:18 i don't have expert 15:18 i have to bless the wands 15:19 Ah. 15:19 Right, Arc can only get Skilled. 15:19 Ok. 15:31 [fh] hothraxxa (Arc Hum Fem Neu), 267242 points, T:50496, ascended https://www.hardfought.org/userdata/h/hothraxxa/fiqhack/dumplog/2017-09-06%2019%3A31%3A26%2C%20hothraxxa-Arc-Hum-Fem-Neu%2C%20ascended.txt 15:31 fh astral is just plain brutal 15:31 anyway i have to go out 15:43 !tell K2 fiqhack updated: fiqhack/4.3.0 8d12e38 Fredrik Ljungdahl: Allow you to dip objects into wonder for properties (but stacking them is dangerous) 15:43 Will do, FIQ! 15:44 k 15:44 K2: Message from FIQ at 2017-09-06 15:43 EDT: fiqhack updated: fiqhack/4.3.0 8d12e38 Fredrik Ljungdahl: Allow you to dip objects into wonder for properties (but stacking them is dangerous) 15:44 FIQhack updated 15:45 here's how it works: 15:46 You can dip objects into them to grant one (more) property. Cursed will remove a property, uncursed will potentially add a redundant one and blessed will ensure that the new one isn't redundant. However, for noncursed potions, there is a 1 - 1 / (existing properties + 1) chance of all properties being lost, and a 1 / (existing properties + 1) chance of any to be added in first place. 15:46 so you can safely add one property, but adding more than one can do nothing or make you lose them instead 15:48 or add a second 15:50 K2: no 15:51 50% of losing properties 15:51 then 50% of nothing happening 15:51 so only 25% of getting a 2nd property 15:51 if it has 2 already, it's 66% and 33% so only 11% 15:51 of adding a 3rd 15:51 ok so, you can add a 2nd or a 3rd 15:51 yes, you can add arbitrarily many 15:51 just the odds are low 15:51 yeah 15:51 just like you can enchant weapons to +50 15:52 GruntHack updated 15:52 just the odds are low 15:52 what changed? 15:52 did you see what we were discussing in the dev channel earlier today? 15:52 no 15:52 fail 15:52 meh 15:53 its a minor cosmetic tweak 15:57 Acid blast -> yellow now 15:58 (In Grunt) 15:59 ahh 15:59 why? it should be green 15:59 like the breath 15:59 ^^ 16:04 FIQ: green was already taken for poison. 16:04 Also, poison breath should be green, acid yellow. 16:04 Like the dragons. 16:05 indeed 16:05 but currently yellow is poison breath 16:05 and green is acid breath 16:05 it's reversed 16:06 but the explosions should stay the same as breath/ray 16:07 -!- LarienTelrunya has quit [Quit: Page closed] 16:08 Well, the poison explosion was already green. That didn't get changed. 16:09 jonadab: so this was the price to pay for more frequent binary checkpoints 16:09 16:09 1504013235_hothraxxa.nhgame 16:09 2017-09-06 15:31 378M 16:09 it's happening way too much 16:10 Yeah, probably need to find a happy medium there. 16:10 Instead of dividing by ten, maybe divide by three? 16:10 [nd] munterplant (Val Hum Fem Law), 1847 points, T:2152, killed by Yeenoghu, while being terrified of a demon 16:10 yeah I'll have to look into it 16:10 and yeah was going to try it 16:10 but why did dividing by ten increase the save size by 50x? 16:11 Apples and oranges. It's comparing ratios. 16:11 Not absolute sizes. 16:11 not really a fan of how it works now 16:11 You could try different logic altogether. 16:11 I think it would be more reasonable to save based on something actually predictable 16:11 but maybe there was a reason 16:11 so I'm just going to change the ratio for now 16:11 jonadab: and thing is 16:12 this was far into a game 16:12 I'd be tempted to say every n turns, but I don't know what n should be. 16:12 it wasn't from turn 1 16:12 Ah. 16:12 so it is going to be even worse 16:12 Not necessarily. 16:12 Because the ratio... oh, wait. 16:12 Hmm. 16:12 It compares the... 16:12 Argh, the logic escapes me right now. 16:13 size of uncompressed save vs size of all the compressed diffs since last checkpoint 16:13 iirc 16:15 Ah. 16:15 Maybe we should compare the size of the _compressed_ save vs the size of the compressed diffs? 16:16 uh 16:16 playing forwars in this replay is *slower* than replaying backwards 16:16 But actually, I tend to think "every n turns" would be a better rule. 16:16 that is not a good sign 16:16 is it checkpointing every single turn? 16:16 That would explain a large filesize. 16:22 there is a checkpoint on line 201425, 201510 and 201600 16:22 that amounts to roughly every 40 turns 16:22 if it kept the same process over the course of the game 16:23 That's probably too often. Especially if it's making the file size asplode. 16:23 Wasn't it previously going for _thousands_ of turns? 16:23 and for quite some time it saved every 1 turn 16:23 Yeah, that's insane. 16:24 If we were going to do that, we wouldn't need the diff engine. 16:24 no wonder playing forwards was slow 16:24 it had to read a full binary save 50 times a time for some time 16:24 when I messed with it 16:25 jonadab: well at least it fixed replaymode being slow :P 16:25 -!- Crawldragon360 has joined #hardfought 16:26 FIQ: Yes, but there's gotta be an in-between setting. 16:26 40 turns too short, 40 _thousand_ turns too long. 16:27 I'd be tempted to try an every 5000 turns rule and see what happens. 16:27 the save rate is weird 16:27 -!- Crawldragon has quit [Disconnected by services] 16:27 -!- Crawldragon360 is now known as Crawldragon 16:28 jonadab: it used to stabilize on around every 10k turns 16:28 I'd try every (turncount/50) turns 16:28 Hmm. 16:29 Maybe. 16:34 !tell K2 pushed a fix that should create checkpoints more rarely 16:34 Will do, FIQ! 16:34 jonadab: now I divide by 2 16:34 checking movecount doesn't neccessarily do the right thing since you don't know when thet last checkpoint was made 16:34 and neutral turnstate doesn't happen every turn 16:35 Oh, you don't know what turn the last checkpoint was, only where it is in the save file? 16:35 yes 16:36 That's... well, I suppose it was designed around the way things currently work. 16:36 let me put it this way 16:36 I don't really understand how the current save system works in detail 16:36 I'm trying to make sense of it as I go 16:36 but it's not easy to understand ais code 16:36 given the way he does things 16:37 He tends to be sufficiently clever to make his code hard for a normal person to maintain, yes. 16:37 I didn't understand his shop auto-ID code either 16:38 and it seems like such a trivial issue 16:39 jonadab: I mean to the point where I have an easier time understanding the nitrohack save code 16:39 which is known to be broken 16:39 I can't _begin_ to understand most of the UI code, but I think that's _at least partly_ due to it being written around the libuncursed (version 1) API, which was designed around the curses API, which is infamous. 16:39 this is probably the biggest reason as to why I am interested in reading the nitrohack save code 16:40 it's fundamentally broken on a theoretical level 16:40 but I have a way bigger chance at making sense of it and fixing its issue than the 4.3 save code 16:40 which I can't hope to understand 16:40 lol 16:40 K2: Message from FIQ at 2017-09-06 16:34 EDT: pushed a fix that should create checkpoints more rarely 16:40 that save is huge 16:40 K2: yeah, sorry 16:40 FWIW I did ask ais before applying the change if it was a good idea :P 16:40 ;) 16:41 FIQhack updated 16:41 It didn't quite have the effect anticipated 16:41 FIQ: Oh, it did. Just... moreso. 16:41 A little too MUCH of the anticipated effect, I would say. 16:41 lol 16:43 jonadab: I'm sometimes tempted to rewrite the 4.3 save system from scratch to something I can understand. The main reason I don't is that it works well enough already and it would be a major task to do 16:43 It works remarkably well, actually. 16:43 jonadab: the thing is, it isn't just replaymode 16:43 I mean, yes, there's some room for adjustments. replaymode, optimization, etc. 16:43 while replaymode is easily the most shafted feature with the 4.3 save system 16:44 there is other issues too that arise from how slow the save system is atm 16:44 like, for example 16:44 when I was debugging hothraxxa's save earlier 16:44 it took 16:44 more than 3 minutes 16:44 to load the save 16:44 in valgrind (around 30s outside) 16:44 it was ridiculous 16:44 to the point where I was tempted to rewind one turn, force it to make a binary save 16:45 and then debug 16:45 Ok, but imagine debugging with the 3.x save system. 16:45 oh 16:46 that's easy 16:46 you can't 16:46 Indeed. 16:46 You have to tell the player, "If you can find a set of steps that consistently reproduces the bug, then I can try to find it." 16:47 Not sure how the NitroHack save system would be for debugging. Better than 3.x, clearly, as long as the save loads. 16:47 it has a completely different issue 16:48 while the save loading is *fast* 16:48 like, *FAST* 16:48 sub-1s to load a full ascension game from turn 1 to asc 16:48 (outside valgrind) 16:48 it has the issue of "hope your save actually loads!" 16:49