00:03:42 Unstable branch on crawl.develz.org updated to: 0.11-a0-818-g331a257 (32) 00:11:12 -!- Blade- has quit [] 00:12:38 -!- MarvinPA has quit [Ping timeout: 240 seconds] 00:20:38 Windows builds of master branch on crawl.develz.org updated to: 0.11-a0-818-g331a257 00:49:22 oh no 00:49:40 I think I have a dirty working copy :( 00:51:05 git stash apply didn't warn me :/ 00:52:55 -!- vadatajs has quit [Ping timeout: 260 seconds] 01:04:14 -!- dtsund has quit [Quit: dtsund] 01:06:22 -!- vadatajs has joined ##crawl-dev 01:23:37 -!- vadatajs has quit [*.net *.split] 01:31:55 -!- dtsund has joined ##crawl-dev 02:13:29 -!- lord-naughty has joined ##crawl-dev 02:22:19 -!- dtsund has quit [Ping timeout: 250 seconds] 02:27:29 -!- lord-naughty has quit [Quit: Leaving.] 02:57:34 -!- vadatajs has joined ##crawl-dev 03:14:46 -!- Xiberia has joined ##crawl-dev 03:25:00 -!- syllogism has joined ##crawl-dev 03:32:08 -!- lord-naughty has joined ##crawl-dev 04:03:19 -!- ainsophyao has quit [Remote host closed the connection] 04:13:14 -!- evilmike has quit [Quit: something happened] 04:50:49 The "keep range" AI could be good for manticores, until they run out of spikes? 04:51:03 Except they're slow moving, I guess.. 04:59:18 @??manticore 04:59:19 manticore (05H) | Speed: 7 | HD: 9 | Health: 36-65 | AC/EV: 5/7 | Damage: 14, 8, 8 | Flags: !sil | Res: 06magic(36) | Chunks: 07contaminated | XP: 164 | Sp: spikes volley (2d10). 05:00:07 OSX builds of master branch on crawl.develz.org updated to: 0.11-a0-818-g331a257 05:03:02 also I'm not sure the spikes have much range 05:03:36 range 6 I guess 05:24:24 -!- mikee_ has quit [Quit: HydraIRC -> http://www.hydrairc.com <- s0 d4Mn l33t |t'z 5c4rY!] 05:26:36 -!- Pingas has joined ##crawl-dev 05:36:27 -!- monqy has quit [Quit: hello] 06:28:47 -!- edlothiol has joined ##crawl-dev 06:32:50 patch making time, nobody panic 06:38:07 which I can't remember how to do... 06:45:33 do you mean git format-patch? 06:45:45 * BlastHardcheese panics 06:49:07 Yeah, sorted it :) 06:49:21 Except the really long filenames XD 06:50:20 Hopefully Xom appreciates being patched 06:52:48 Keskitalo: perhaps we could give manticores a different speed? if they had keep range, limited ammo, and fastmove they'd be like megacentaurs 06:53:38 dire centaur 06:56:38 -!- MarvinPA has joined ##crawl-dev 06:57:51 we still need dire wolf spiders 06:58:14 so, extinct spiders? 06:58:40 yes, for the Tar Pits branch of crypt 06:58:47 Eronarn: Is being a megacentaur good or bad? :) 06:58:56 need rasphyx to enter, the deeper you go the older the fossils are 06:59:21 Could also make fast move, but superslow firing, if that makes sense.. 06:59:28 I assume it will have all the monsters that were removed 07:00:31 Keskitalo: i was thinking fastmove 150, attspeed 100, and make sure their spikes are doing noticeably more damage than their melee 07:00:36 so that there's an incentive to try to 'rush' them 07:01:33 we don't want keep range on centaurs, that would be mean, but something like the mythical 'centaur with better AI' would be a good rarer monster 07:05:58 -!- Ragdoll has joined ##crawl-dev 07:06:39 -!- Pacra has quit [Quit: This computer has gone to sleep] 07:06:53 Yeah, I think that sounds good. Could have even a slower firing speed, so you would catch up with them more when you fire. You'd have a situation where if they'd fire twice in a row, you could catch up with them. 07:07:45 Keskitalo: reading the summary evilmike did... for the reverse trample, i had that idea a while ago for a summon spell 07:07:54 Might make more sense on a different monster. Like a "turret turtle", which keeps range, but withdraws to shell to fire, and stays there for a bit. 07:08:10 smite target a creature, and out of a corner, smoke starts to appear. a few rounds later, a really powerful constricting, reverse-trampling X appears 07:08:33 it grabs the enemy and pulls it away from combat to maul it 1-on-1 07:08:57 spell only works if there's a nearby corner 07:09:11 lol 07:09:31 it would also work well on a spider that grabs you and drags you into webs 07:09:35 interesting tactical considerations there 07:09:42 perhaps this is how we could implement trapdoor spiders 07:10:02 -!- ToBeFree has joined ##crawl-dev 07:10:03 they don't do that though 07:10:16 yeah, but we're not going to have them move the player to another lvel 07:10:16 Eronarn: Oh yes, that sounds good for trapdoor spiders - they could attempt to pull you to webs. 07:10:27 With crocodiles/alligators, they could attempt to pull you to water. 07:10:37 yeah 07:10:51 give it to krakens if they eventually get constrict, of course 07:11:09 kraken tentacles could get the pull too, yeah 07:11:36 this sounds like something we might want to have the bitmasked AF flavors for though 07:11:42 The summon idea sounds nice as well; more "functional" summons would be so good to have. 07:12:09 Keskitalo: yeah, for this version one of my top goals is to shape up some of the bad spell schools 07:12:17 Hmm, that tentacle summon spell could use pull as well. 07:12:37 as i see it: summ, hex, tmut, earth (only bad in how boring the effects are) 07:12:50 The pinning idea sounded good for iron shot/LCS. 07:13:19 Keskitalo: though, it may make more sense for icicle, since it can melt to end the effect 07:13:49 and if we added it to icicle we could drop the damage, so that the imb vs. icicle choice is a little more interesting than 'do i have ice magic high enough' 07:14:08 do monsters heal much? 07:14:21 Yup. The push-back idea for IMB sounded reasonable as well. 07:14:39 knockback on a spell that deals damage is risky 07:14:51 see also imb smoketrails 07:15:00 I don't think it was meant to be guaranteed though.. 07:15:12 was thinking a pois/earth spell called Salt Wounds, which blocked healing 07:15:43 Monsters healing isn't generally tactically very relevant.. 07:15:56 drat 07:15:57 yeah, that's more of a strategic thing... stuff does heal while you're on the level, but crawl fights are short 07:16:14 In my last game I noticed it only when I was cutting off hydra heads. :) 07:16:15 it might be better as a secondary effect of another spell 07:16:16 oh well 07:16:31 There's also already a healing-blocking effect, Sick. 07:16:37 eronarn: knockback on a spell that deals damage seems okay if the knockback chance isn't that high 07:17:02 elliptic: if you can't make the chance scale to be reliable, that's fine, yeah 07:17:38 like, 50% max (at close range against small monsters) or something like that 07:18:11 elliptic: what if it worked better at long ranges rather than short? to parallel iood 07:18:25 eronarn: I was thinking it would be better for IMB to complement IOOD 07:18:28 so the reverse 07:18:45 knockback then iood combo 07:19:12 fire iood, then imb them back so the iood has enough time to catch up 07:19:21 specifically, I was thinking that IMB should have full or near-full range, but be stronger in some way at close range 07:19:50 not sure that knockback is the most interesting effect but I might try it out at some point 07:19:59 you aren't going to be able to actually get someone to move away from you with knockbacks, generally, unless they're a caster or something 07:20:06 but you can make them not get closer 07:20:25 so you'd want to be casting it while they're already at range from you 07:20:37 (against non-spellcasters, knockback in melee plays out a lot like stun) 07:20:47 is there some sort of pull spell 07:20:47 ? 07:21:07 HousePet: apportation :) 07:21:13 (it doesn't work on monsters) 07:21:23 unless they are unknown mimics! 07:21:34 tome has a few really cool spells that deal with moving enemies around 07:21:36 i tried an apportation on a monster hoping it would disarm them :/ 07:21:46 one is a spell that grabs an enemy and pins them to the ground 07:22:09 another is a spell where you choose a target, and anything around you is swapped with anything around the target, and it deals a debuff to anything that was swapped except for you 07:22:14 attacks that move enemies around have the potential to be really broken, unfortunately 07:22:20 the latter effect is just really, really awesomely fun 07:23:02 some sort of swap could be interesting, yes 07:23:29 i liked the disarm spell in arcanum :( 07:25:27 HousePet: for a while i've wanted a 'baleful apportation' spell: can grab weapons, limbs, etc. 07:25:52 ouch 07:27:29 more powerful apportations for disarming and moving creatures to melee range would be fun 07:27:39 Eronarn: Another "pull AI" idea: pack monsters that have pull try to pull their foe out on the open, near their comrades 07:27:44 canines come to mind :) 07:31:49 One more high-level non-evil summon idea: Stampede. You cast it, choose a direction, and then a herd of stampeding (neutral) elephants starts appearing a few at a time from the opposite direction towards to where you want them to stampede. Like a wave of monsters, pushing everything in their path. They only attack stuff that gets in their way. 07:33:09 awesome :) 07:36:59 Keskitalo: i love the idea but it might need code to make monsters move out of the way if something neutral has trampled them and could trample them again 07:37:10 otherwise a single elephant will shove something all the way across the screen 07:37:19 Only with high enough intelligence of course. :) 07:37:21 there's also the case of ooY 07:37:29 -!- ghallberg has quit [Remote host closed the connection] 07:37:29 right now the elephant will just keep attacking 07:37:42 i already proposed being able to make it shove the orc off to one side or another 07:38:00 this wouldn't be too hard, but it would be a big behavior difference for trample 07:38:21 That's bad tactically for the trampler though ;) 07:38:51 I actually think there should be collision damage from being trampled to walls, and why not monsters.. it's just that, what to do with water and lava 07:39:39 Keskitalo: think of it like the orc diving out of the way, to one side, rather than being slammed into something 07:39:51 -!- edlothiol has quit [Ping timeout: 265 seconds] 07:39:57 collisions would only happen if there's nowhere to actually be moved to 07:40:46 Summon Rolling Boulder! 07:41:07 HousePet: Too close to IOOD imho, unfortunately :) 07:41:40 Eronarn: Nah, the orcs should only make such moves if they press hjklyubn ;) 07:41:41 sigh 07:42:26 I blame the existance of conj, sum and translocation schools 07:42:49 Keskitalo: such 'easy' chain trampling can create issues, i fear 07:43:14 HousePet: tmut, rolling boulder form 07:43:16 Of what kind? 07:43:19 create a bould from nothing, summon a boulder from elsewhere, grab nearby boulder and make it roll 07:44:02 tmut version would be funny if you can't change direction until the poly ends :D 07:46:04 do not use anywhere sudden lava could appear. eg abyss 07:52:53 -!- HousePet has quit [Quit: Page closed] 07:54:08 -!- vadatajs has quit [Ping timeout: 240 seconds] 08:07:28 -!- vadatajs has joined ##crawl-dev 08:11:38 -!- st_ has joined ##crawl-dev 08:19:00 Have we had recent major GUI changes in Tiles, or are we expecting any? 08:19:46 Because I'm reading the KiSS usability project report, and they seem to be under such impression. I'm a bit baffled, but it may be that I've given such an impression. 08:19:57 I certainly don't recollect such changes.. 08:20:13 no one even maintains local tiles as far as I know 08:22:36 "Most of the results indicated major issues with the GUI of DCSS which was however up to change based on the communication with the development group and was thus ignored as an notable issue." 08:22:40 Sigh. 08:22:55 I have no idea what I've said. :P 08:26:57 Well, the report still looks pretty good, with lists of problems both from the team's own evaluation and from the user testing. 08:27:29 I haven't looked at the tutorial redesigns yet, but it sounds like they've shortened them a nice bit and suggest they could still be shortened. 08:30:37 -!- Dixbert has quit [Read error: Connection reset by peer] 08:30:54 -!- valrus has joined ##crawl-dev 08:46:47 -!- edlothiol has joined ##crawl-dev 08:53:37 -!- valrus has quit [Remote host closed the connection] 09:26:12 -!- edlothiol has quit [Quit: rcirc on GNU Emacs 23.3.1] 09:51:12 how about preventing the v/8 guards from taking stairs? 09:51:13 galehar: You have 1 message. Use !messages to read it. 09:51:17 !messages 09:51:20 (1/1) nfogravity said (11h 53m 38s ago): i have a working fsim skeleton. it doesn't have every function of the old one, and doesn't output to a file, but i'll make a patch and upload it to mantis tomorrow for you to play with 09:51:27 good 09:53:46 -!- nfogravity has joined ##crawl-dev 09:54:27 galehar: Sounds good, I think that's been discussed in a SourceForge FR even 09:54:38 By which I mean the idea has been around for a while indeed 09:55:15 hey galehar, i'm going to finish cleaning up my code today and upload what i have 09:55:29 nfogravity: fsim improvements! <3 09:55:49 yeah :D 09:56:13 nfogravity: good, thanks 09:56:33 Keskitalo, http://i.imgur.com/98Bst.png 09:59:29 this is a nice feature, but personnally, I'm more interested on seing how damage scale with skill 09:59:40 i know 10:00:06 I'm also interested in cleaning up the code, so if you've started it, I want to use your code 10:00:19 please do, i rewrote it completely 10:00:33 for example, there's no reason to pass the missile_slot through every functions. Check the quiver instead. 10:00:51 already done 10:00:57 good 10:01:02 it's a massive massive rewrite 10:01:23 well, that was probably the thing to do! 10:01:40 ok have to go now 10:02:14 I'm still looking for a formula to scale mimics hd and damage with depth. I'm taking suggestions. 10:02:20 'later 10:03:07 right now i'm trying to figure out what unwind_var does 10:13:47 elven dagger tiles not centered within tile (https://crawl.develz.org/mantis/view.php?id=5512) by nicolae 10:29:02 -!- ainsophyao has joined ##crawl-dev 10:29:15 -!- edlothiol has joined ##crawl-dev 10:57:25 is a lerp too simple for the mimic scaling? 11:13:22 -!- ainsophy_ has joined ##crawl-dev 11:13:25 -!- ainsophyao has quit [Read error: Connection reset by peer] 11:55:38 Unstable branch on crawl.akrasiac.org updated to: 0.11-a0-818-g331a257 (32) 11:55:52 oh hey that script is still running, awesome 12:18:14 -!- lord-naughty has left ##crawl-dev 12:21:00 galehar: if you want to be really mean, have them not follow up stairs and re-align themselves around the stairs so that the ones with polearms are in reach range and the others move inside 12:21:01 getting hit by 20+ guards in one turn should make people stop stairscumming them 12:21:54 actually, that wasn't serious but it makes me think that maybe now that reaching is for all polearms, the v8 stairs should be closer together 12:29:22 ??poison resistance 12:29:22 poison resistance[1/2]: Makes you immune to normal poison (this includes chunks!), damage from normal poison attacks, and confusion from mephitic clouds. Halves damage and poison from Poison Arrow. Recommended for Swamp and Snake, of rather little use elsewhere. Only 90% effective against poisonous attacks in 0.10. 12:31:42 stair dancing them is usually a worse idea than ctele or blinking into one of the corridors so it wouldn't really change that much 12:33:39 "usually" should be "always" though 12:34:54 -!- MarvinPA has quit [Read error: No route to host] 12:35:18 -!- MarvinPA has joined ##crawl-dev 12:36:30 -!- MarvinPA has quit [Read error: Connection reset by peer] 12:36:51 -!- MarvinPA has joined ##crawl-dev 12:37:06 st_: although, i also don't think it's a good thing if ctele is automatically the "best" solution either 12:43:18 i think it'd probably be more fun if it blocked ctele, or even tele 12:44:01 actually, here's an idea: apply subvaults to the v8 entrance, and have different (hard) defenses each time 12:44:35 Eronarn: I pretty much just suggested this on the tavern thread 12:44:41 with A) each weak against certain tactics and strong against others B) penalizing stairscumming by having the guards able to pile on the stairs if you do it 12:44:54 replace all the vault guards with mara (but only one is real) 12:44:56 Each upstairs/downstairs could be a subvault, increasing in difficulty the further down you are 12:45:31 mumra: that'd work well with the vault boxes idea 12:45:51 exactly 12:46:22 each box can get populated by a subvault or a random set of monsters from a weighted selection 12:47:04 my thought for the vault guard defenses would be that we could have stuff like relevant statues, or a moat 12:47:45 and have some upstairs/downstairs vaults as well with a higher chance than normal to get used - so you'll only occasionally get stairs in corridors (just to keep things a bit random) 12:48:46 it'd be nice if someone did this for next version :) 12:50:21 yes ... someone ... ;) 12:51:09 well the subvault / level building part isn't actually that bad, a lot of infrastructure is there already 12:51:27 it'll just need as much content as possible from vault designers 13:16:50 -!- neunon has joined ##crawl-dev 13:16:50 -!- neunon has quit [Changing host] 13:16:50 -!- neunon has joined ##crawl-dev 13:16:50 -!- Dixie has joined ##crawl-dev 13:30:46 -!- lord-naughty has joined ##crawl-dev 13:40:40 -!- lord-naughty has quit [Quit: Leaving.] 13:53:37 I honestly don't think there is any issue with people stairscumming v:8... it is quite dangerous with most chars 13:54:15 making the vault guards more interesting is good but the way to do this isn't to remove non-dominating strategies... good to have many different strategies for them 13:55:58 the vault pretty much forces that... or a teleport right at the start 13:56:11 or just running past them, or blinking past 13:56:29 which are usually better options than stairscumming, as st_ said 13:56:48 they rarely can be ran past, blink indeed works but is way riskier 13:56:57 s/rarely/always/ 13:57:09 seriously, it is guaranteed to be able to run past them if you are hasted 13:57:27 stairscumming means you get a few every iteration, can deal with them easily them heal up 13:57:40 stairscumming means you die unless you have really great defenses 13:57:46 I've watched many people do this 13:57:53 yeah, in practice stairscumming v:8 is a terrible idea 13:58:13 I'm a terrible player, yet I never had even a hint of danger stairscumming V:8 13:58:28 every time you go downstairs you have to deal with the fact that a titan might have walked into view, or a shadow dragon, or an ancient lich 13:59:06 every turn you spend there you have to deal with the fact that a titan may walk into view, or... 13:59:07 one time I walked downstairs while stairscumming (after killing most of the guards) and took 150 damage before I got an action 13:59:25 a titan/quicksilver dragon/other stuff combination 13:59:29 giving 8 enemies the chance of a simultaneous free hit on you tends to be a bad idea 13:59:57 one time i went downstairs and got tormented and lost 150 hp. what's the difference? 14:00:01 I do stairscum sometimes, when I have limited haste and good defenses (i.e. many trog chars) 14:00:10 zannick: ? 14:00:16 sounds like a xl12 char with 0 AC 0 EV 14:00:34 kilobyte: no 14:00:36 that's ridiculous, kilobyte. i've lost several very good characters 14:00:48 on the other hand, it's staying there that's decidedly bad for your health 14:01:35 !lg * killer=vault guard s=place 14:01:35 198 games for * (killer=vault guard): 85x Vault:7, 50x Vault:8, 13x D:11, 12x D:14, 9x D:10, 6x D:15, 5x D:16, 5x D:17, 4x D:12, 4x D:19, 3x D:18, Vault:4, Vault:3 14:01:38 kilobyte: ^^ 14:01:47 vault:7, the most common place to die to a vault guard 14:02:06 this is pretty telling IMO 14:02:11 !lg * killer=vault guard place=vault:7 s=xl 14:02:11 85 games for * (killer=vault guard place=vault:7): 16x 20, 11x 21, 11x 22, 9x 17, 8x 23, 7x 19, 5x 24, 5x 18, 3x 25, 2x 26, 2x 27, 2x 15, 2x 13, 16, 14 14:02:31 !lg * killer=vault guard place=vault:7 s=class 14:02:31 85 games for * (killer=vault guard place=vault:7): 15x Berserker, 8x Wizard, 7x Fighter, 6x Enchanter, 6x Transmuter, 5x Necromancer, 5x Conjurer, 4x Fire Elementalist, 4x Priest, 3x Venom Mage, 3x Air Elementalist, 3x Summoner, 2x Assassin, 2x Crusader, Earth Elementalist, Stalker, Abyssal Knight, Paladin, Chaos Knight, Gladiator, Artificer, Wanderer, Hunter, Monk, Ice Elementalist, Healer 14:02:50 you rarely die while pulling them, you usually die while actually trying to dispatch them after pulling 14:03:11 yeah, dying while pulling them would count as v:8, no? 14:03:25 at least most of the time, so far as i recall 14:04:09 people don't usually die to them while going upstairs, no 14:04:10 you start at full hp every iteration so you'd need paralysis or terrible luck 14:04:37 they die because they are surrounded by eight of them and injured 14:04:45 -!- heteroy has joined ##crawl-dev 14:05:01 oh, i usually try to choose staircases next to walls :) 14:05:04 vs being surrounded by ~30 of them, plus shadow dragons, plus titans, and injured 14:05:17 is there anything new in trunk that is specifically wanted to be playtested? 14:05:22 kilobyte: what part of everyone telling you that you run past them or blink past them do you not understand 14:05:25 heteroy: Spider 14:05:45 seriously, nobody is advising just sitting in the center of v:8 killing everything there, unless you have tornado or refrigeration or something 14:05:49 k. anything race/class/god-wise? 14:06:07 the point is that there are many options and most good players seem to prefer not to stairscum 14:06:13 the titan next door, in a nasty place with no retreat 14:07:36 as I said at start, making vault guards do something interesting would be cool... but basing changes on some perceived need to address stairscumming them is misguided when more people die to them on v:7 than v:8 14:08:44 far more people dying on v:7 than v:8 is a clear proof of stairscumming being the dominant strategy, since no other way would produce a death there 14:08:53 it doesn't prove it's the most successful one, yeah 14:08:57 just most used 14:09:06 it doesn't prove that, what are you talking about? 14:09:14 these are just deaths to vault guards 14:09:23 when stairscumming, you don't die on V:8, you die on v:7 14:09:27 i'd also guess that people die on v:8 to nonvault guards after being trounced by vault guards 14:10:11 kilobyte: when you run past the vault guards, you don't die to them at all 14:10:33 the point is to not stand and fight the ambush at all, yeah 14:10:47 this is also good advice for the other ambush vaults 14:10:57 like snake:5 ambush, or evilmike ambush 14:11:03 run/teleport away 14:11:07 it is very hard to automatically classify games as deaths due to particular methods of tackling v:8 14:11:29 zannick: sure, hard to determine much from the statistics... but what we can tell is that stairscumming has its risks 14:11:33 this sounds like another case of "badwiki misinformation" more than anything perpetrating the bad strategy 14:11:40 if you go past them, you instead die to any other of depth:48-58 monster 14:12:04 or you kill them all from a much better position, get the rune, and win 14:12:15 personally, with squishy wizard characters, i've found stair pulling to be very effective. i've mostly# died on V:8 in the quadrants. 14:12:59 the corridors between quadrants are pretty safe, compared to the alternative of getting completely surrounded and being completely surrounded again when re-entering the level 14:13:18 it just doesn't feel like anything close to a dominant strategy to me, and plenty of other people seem to agree 14:14:00 !lg . place=Vault:8 s=killer 14:14:00 5 games for kilobyte (place=Vault:8): a golden dragon, a titan, an ancient lich, an iron dragon, a quicksilver dragon 14:14:02 !lg . place=Vault:7 s=killer 14:14:03 2 games for kilobyte (place=Vault:7): a lich, a stone giant simulacrum 14:14:04 so our focus shouldn't be on nerfing it, but rather making vault guards more interesting in general, regardless of your strategy 14:16:05 !lg . place=v:8 s=killer 14:16:06 2 games for Zannick (place=v:8): a yaktaur captain, a stone giant 14:16:34 so embarrassing, such non-ood monsters :) 14:16:55 !lg * place=v:8 s=killer 14:16:55 980 games for * (place=v:8): 233x a stone giant, 100x a yaktaur captain, 82x a titan, 59x a shadow dragon, 50x a vault guard, 41x a quicksilver dragon, 32x a storm dragon, 31x a frost giant, 28x , 27x an ancient lich, 26x an Executioner, 24x a fire giant, 24x an ettin, 20x a tentacled monstrosity, 17x a golden dragon, 16x a lich, 14x an iron dragon, 9x a deep elf annihilator, 6x a yaktaur, 5x a Ba... 14:17:10 Quote from the badwiki: "For melee characters, the best way to cope is to lure a handful of guards upstairs so that you need not fight quite so many of them at once, kill them, then rest and repeat." 14:17:12 zannick: pretty typical though 14:17:19 This is why I argue that the badwiki should just be shut down 14:17:24 elliptic: yeah 14:17:46 !lg * place=v:8 killer=tentacled monstrosity s=cv 14:17:46 20 games for * (place=v:8 killer=tentacled monstrosity): 11x 0.10, 4x 0.10-a, 3x 0.11-a, 0.9, 0.4 14:18:34 constriction: pretty deadly 14:18:40 yes 14:18:46 especially while stairscumming :P 14:18:46 speaking of tentacled monstrosities, they're thoroughly insane now 14:19:19 constriction formulas need to be scrapped and rebuilt from scratch I think 14:19:55 maybe damage is okay but success chance and chance of pulling away are pretty crazy 14:20:08 since they sometimes spawn in Swamp, are a player summon, and random pan lords with it tend to summon eight or so before you even walk up to them, I'd greatly nerf them rather than make them rarer 14:20:31 some other monster could have insane constriction if needed, but not a popular summon 14:20:41 elliptic: formulas? I say the same about code. 14:20:43 SHT was already overpowered, yes 14:21:27 though arguably most summons are 14:21:28 holy cow, they summon 3-5 _per cast_ 14:21:38 (monsters) 14:21:50 mainly SHT summons large aboms when players cast it 14:21:52 -!- alefury has joined ##crawl-dev 14:22:20 also Mnoleg needed the buff :P 14:22:25 but yeah 14:24:57 ghrmblahdammit, gcc-4.7 final still dies with an ICE when building Crawl with LTO :( 14:25:05 so does gcc trunk 14:26:37 wow, i was gone for almost a week and there still is no 0.10.1 news post 14:26:43 !tell SamB do you have any clue debugging/properly reporting bugs on gcc? Both 4.7 and trunk die with an ICE when building Crawl with LTO=y. 14:26:43 kilobyte: OK, I'll let SamB know. 14:27:02 (SamB is likely to have a clue, he maintains gcc-doc in Debian) 14:27:12 kilobyte: not yet I don't ... 14:27:12 SamB: You have 2 messages. Use !messages to read them. 14:27:20 !messages 14:27:20 (1/2) Jevouse said (12w 2d 1h 18m 32s ago): you're a nice person to fix stuff for us endusers 14:27:36 !messages 14:27:37 (1/1) kilobyte said (53s ago): do you have any clue debugging/properly reporting bugs on gcc? Both 4.7 and trunk die with an ICE when building Crawl with LTO=y. 14:27:45 SamB: oh, you're here, cool. 14:28:58 I wasted about an hour (counting human time, not waiting for builds to fail), and couldn't get even to building gcc from upstream sources (they immediately reject reports for sources obtained elsewhere) 14:28:59 well, obviously it's best if you can reduce the size of the test case ;-) 14:29:14 -!- Pacra has joined ##crawl-dev 14:29:26 oh, really? 14:29:54 all of docs how to reduce test cases apply to one-file compiles, not to LTO on massive projects 14:30:04 and Crawl is not exactly modular :( 14:30:17 well, obviously that's not easy to give rules for 14:30:29 -!- MarvinPA has quit [Read error: Connection reset by peer] 14:30:47 -!- Dixie has quit [Read error: Connection reset by peer] 14:31:29 -!- MarvinPA has joined ##crawl-dev 14:31:50 -!- lord-naughty has joined ##crawl-dev 14:32:51 oh, there might be something on the wiki about reporting LTO bugs? 14:36:49 kilobyte: so they really dismiss out-of-hand *any* reports involving vendor versions? 14:37:10 I don't see anything about that on http://gcc.gnu.org/bugs/ 14:38:12 at least I heard complaints that they tend to for Debian versions, which are indeed very heavily patched 14:38:55 * SamB isn't sure they're as heavily patched as you might think 14:39:24 but there *is* certainly some lack of transparency to the patching 14:40:12 (There are a lot of conditionals in rules.patch) 14:42:20 115 patches, quite a bunch above 50KB each 14:42:22 I guess you should perhaps start with bugs.debian.org ? 14:43:43 I don't think many of those are actually used for any given build 14:44:40 hmm 14:44:50 naesten@hydrogen:~/hacking/debian/gcc-doc% zgrep '^[^ ].*:$' /usr/share/doc/gcc-4.6/README.Debian.i386.gz | wc 14:44:50 66 66 1078 14:44:50 n 14:45:17 Qwertycoatl (L9 NaTm) (D:9) 14:46:24 kilobyte: it seems that unrands aren't currently autoinscribing 14:46:41 kilobyte: so, did you try bugs.debian.org yet? 14:47:46 SamB: as in, throwing the bug onto Matthias Klose and co's shoulders? 14:48:15 well, yes, it does sound kind of silly when you put it like that 14:48:39 I didn't, they seem to be massively overworked 14:48:56 but you have a point -- it can't hurt to at least communicate 14:49:18 they might at least be able to offer some suggestions about how to proceed 14:49:28 I'm a mere user when it comes to gcc internals, yeah 15:06:00 -!- mumra has quit [Quit: offski] 15:45:29 did hexes and charms crosstrain at some point? it seems like something that would be appropriate. 15:46:23 elliptic: ghallberg put a proposal for a new formula on the wiki. Since the current one is so bad, there couldn't be much harm in trying this one 15:46:44 I'm talking about evading constriction btw 15:46:53 allllmost done, galehar 15:47:12 i promise it's worth it 15:48:14 :) 15:49:34 heteroy: they did, at a 100% ratio. 0% since then. 15:49:41 -!- monqy has joined ##crawl-dev 15:50:16 -!- heteroy_ has joined ##crawl-dev 15:50:18 changing the escape formula seems really important, yeah 15:50:56 would be hard for the proposed one to be worse :P 15:51:19 the proposal is from galefury btw 15:51:44 How about this formula (assuming the player is trying to escape, swap str and HD for monster escape attempts): 15:51:47 x_chance_in_y(5+n*(3+max(0,2+div_rand_round(2*str-HD,5))),30+3*(attacker_size-defender_size)) 15:51:50 with n the number of escape attempts, starting at 1. This formula gives a fairly high chance to escape early constrictors, but only high strength chars have a high probability to quickly escape late game constrictors. 15:51:53 -!- heteroy has quit [Ping timeout: 246 seconds] 15:51:56 -!- heteroy_ is now known as heteroy 15:52:09 oups 15:52:15 Revised version: x_chance_in_y(5+n*(3+max(0,2+div_rand_round(2*str-HD,5))),35) 15:52:18 Examples: https://docs.google.com/spreadsheet/ccc?key=0Aj0ZoxMMMg-6dG80ODFPRFdTSkVaTk5pY3hQQmxxV2c 15:52:36 why str and not dex? or has that been discussed before? 15:53:38 maybe there: https://crawl.develz.org/wiki/doku.php?id=dcss:brainstorm:effect:constriction 15:54:03 dex/ev should help you not get hit in the first place, presumably 15:54:04 that formula has insufficient randomization 15:54:26 -!- ainsophy_ has quit [Remote host closed the connection] 15:54:57 specifically, you are guaranteed to be able to escape by a specific number of turns, and you can compute how many turns that is ahead of time if you know the formula 15:55:39 should become more and more likely but never guaranteed, right? 15:55:49 and not piecewise ;) 15:56:11 <|amethyst> I don't know if it needs to become more likely after a point 15:56:31 though maybe it is okay since the rate of increase of escape chance doesn't look that high 15:57:55 I guess it is for 20 strength chars though 15:58:27 personally I'm not really sure why this should depend on n at all 15:58:42 -!- ghallberg has joined ##crawl-dev 15:58:49 fyi, formula for nets: x_chance_in_y(40 - you.stat(STAT_STR), 66) 15:58:57 I think having str there at all is bad: constriction done by the player could use str, yeah -- but when it's a monster that's constricting, you'll have tentacled monstrosity splat everyone but characters who already could take being constricted for a while 15:59:15 I joined ##crawl instead och ##crawl-dev. Thought there was loads going on tonight... 15:59:21 which is the change to *not* escape 15:59:26 *chance 15:59:35 galehar: I'm playing diplomacy with some friends from school, I rolled France and I'm kicking ass. 15:59:55 :) 16:00:13 ghallberg: such an unrealistic game! albinoblacksheep brought us the proof :) 16:00:15 Well at least I was for a while :) 16:00:26 Now the germans and ryssians are pretty much crushing us. 16:00:27 kilobyte: maybe, yeah... I wonder how well just using a fixed escape chance that depends on monster HD and nothing else would work 16:01:11 kilobyte: well, net use str, so at least it is consistent if constriction does too 16:01:16 and it makes sense 16:01:42 also, players with no str are likely to have a blinking source at the depth of tentacled monstrosity 16:01:53 also I want to propose stunning the monster for a fraction of a turn (maybe half a turn or so) when you successfully escape 16:02:02 or some other mean of not getting in melee range 16:02:05 to prevent randomized energy from immediately letting it re-constrict 16:02:18 elliptic: goo idea 16:04:26 I think a fixed chance could work too. Statistics say you'll eventually get out ;) 16:04:34 and it's much simpler 16:04:57 dependence on n does mean that your turn isn't completely wasted if you fail though, which might be good 16:07:08 I think I might go for a formula of the form: chance of failure = (linear function of HD)/(linear function of n) 16:08:03 could be something like (a - b/n) to make it tends to a. And make a depend on HD and str. 16:09:52 i'd make a increase with dex as well, as kilobyte said before, it would cause TM to splat everyone but people who could already take some hits 16:10:12 "it" = using str and not dex 16:10:16 I guess with that sort of formula you have to be careful to make sure you don't make the success chance zero sometimes 16:10:41 zannick: I'd rather use neither str nor dex, for simplicity, at least at first 16:10:48 can try to add those in later maybe 16:10:56 elliptic: that's another option 16:11:14 is there consensus on cutting 3 levels from V? 16:11:50 I haven't heard anyone argue for a V length of longer than 5 levels, yeah 16:12:17 does that significantly mess with the xp curve? 16:12:25 is that accompanied by moving V down 3 floors on average? 16:13:41 (to preserve the right monster mix for "v:8") 16:13:59 Zannick: we certainly want to keeps v:8 depth 16:14:38 but if we wait for portal_branches, we can set it's depth independtly from the depth of its entry 16:14:53 ah, well, then 16:15:31 we can still do it now, move its entry (and break save compat) and revise depths on the merging of portal_branches 16:16:42 nfogravity: I've put some XP stats on the wiki: https://crawl.develz.org/wiki/doku.php?id=dcss:brainstorm:dungeon:layout 16:16:50 the answer is yes and that's the point 16:17:50 there's also crypt who is a prime candidate. Worth a lot of XP, not much threatening and very boring. 16:18:12 but ghoul bands! :/ 16:18:19 fr: make ghouls fast :D 16:18:25 it's been proposed to remove it and use the endings as a portal vault 16:18:48 where does tomb go? 16:18:53 that'd presumably need rebalancing 16:19:23 -!- dtsund has joined ##crawl-dev 16:19:34