00:06 -!- Menche has quit [Remote host closed the connection] 00:06 -!- Menche has joined #hardfought 00:32 [nh4] Jim (jimbentler) (Tou Hum Mal Neu), 13240 points, T:527, killed by a gnome 01:13 -!- raisse has joined #hardfought 01:13 -!- mode/#hardfought [+v raisse] by ChanServ 01:31 -!- Grasshopper has quit [Ping timeout: 248 seconds] 01:41 -!- hpardis has quit [Ping timeout: 268 seconds] 01:44 -!- noty has quit [Quit: WeeChat 1.9.1] 01:59 [nd] Raisse (Arc Dwa Fem Law) had Grayswandir bestowed upon her by Quetzalcoatl, on T:3393 02:06 [nd] Raisse (Arc Dwa Fem Law) became literate by reading a scroll, on T:4512 02:52 -!- hpardis has joined #hardfought 03:00 -!- hpardis has quit [Quit: Leaving] 03:25 -!- shpingrlde is now known as flump 03:25 !setmintc 03:25 flump: Min reported turncount for flump removed. 03:26 -!- flump is now known as shpingrlde 03:26 [nd] Raisse (Arc Dwa Fem Law) had Snickersnee bestowed upon her by Quetzalcoatl, on T:2941 03:28 [nd] Raisse (Arc Dwa Fem Law) consumed animal products for the first time, by eating a coyote corpse, on T:4139 03:30 [nd] Raisse (Arc Dwa Fem Law) had Trollsbane bestowed upon her by Quetzalcoatl, on T:5137 03:30 quetz is trolling me 03:33 snick's not too bad? 03:33 [nd] Raisse (Arc Dwa Fem Law) became literate by engraving "gfj", on T:6668 03:34 [nd] Raisse (Arc Dwa Fem Law) had Demonbane bestowed upon her by Quetzalcoatl, on T:6959 03:35 more banes, thanks coaty D: 03:51 now I've zapped myself with a wand of invis because I thought it was a nothing-happens wand 03:51 meaning I can't sell the useless artifacts to the general store any more until the RNG sends me a mummy 03:53 oh! I know what happened, I got blinded by rotten food between scribbling in the dust and trying the wand 03:53 so I didn't see the engraving vanish 03:54 but I still saw the monsters by telepathy so didn't notice I was blind 03:56 heh, i've tried to buc-test while blind a few times. not sure about engrave-test, but probably 03:57 some things ID even while blind (like digging, you can feel the gravel) 04:14 watching a zen game would be fun 04:24 -!- Grasshopper has joined #hardfought 04:59 -!- aoei has joined #hardfought 05:03 -!- Grasshopper has quit [Ping timeout: 248 seconds] 05:20 -!- LarienTelrunya has joined #hardfought 05:26 [nd] ShivanHunter (Wiz Elf Mal Cha), 169 points, T:508, killed by a small mimic 05:29 ♥ 05:29 LarienTelrunya: Message from K2 at 2017-12-01 20:51 EST: slex has been updated on hardfought - 2.1.1 branch 05:30 !tell K2 awesome! thanks! 05:30 Will do, LarienTelrunya! 05:30 [nd] ShivanHunter (Wiz Elf Mal Cha), 267 points, T:937, killed by a bat 05:39 [nd] ShivanHunter (Wiz Elf Fem Cha), 937 points, T:1909, killed by an iguana 05:46 [nd] Raisse (Arc Dwa Fem Law) had Sunsword bestowed upon her by Quetzalcoatl, on T:2649 05:49 !who 05:49 LarienTelrunya: Terumasa [nd] ShivanHunter [nd] Raisse [nd] mahershalal [nd] 05:49 [nd] ShivanHunter (Wiz Elf Mal Cha), 1584 points, T:2056, killed by a gray unicorn 05:53 [nd] mahershalal (Val Hum Fem Law), 174 points, T:373, killed by a water moccasin 05:58 [nd] ShivanHunter (Wiz Elf Mal Cha), 345 points, T:724, killed by a gnome 06:05 [slex] AmyBSOD (Fem Ang Fem Cha) shouted "testshout", on T:4722 06:05 okay so that works :) 06:06 [nd] ShivanHunter (Wiz Elf Mal Cha), 812 points, T:1506, killed by a small mimic 06:09 [nd] ShivanHunter (Wiz Elf Mal Cha), 185 points, T:546, killed by a fox 06:19 -!- ais523 has joined #hardfought 06:19 -!- ais523 has quit [Changing host] 06:19 -!- ais523 has joined #hardfought 06:21 Beholder: messages? 06:21 ais523: Message from FIQ at 2017-12-01 05:26 EST: 90405235e5037d033b90af0af967d290f62225cc does PatR know about #adjust a a ? it should really be made into the only way to merge things IMO, the new way he added to *not* merge when adjusting to a different letter is IMO much more clumsy 06:21 ais523: Message from FIQ at 2017-12-01 05:27 EST: I was not sure if I should send a mail with this complaint or not 06:22 FIQ: do it, I'm pretty sure that self-adjust is a bug (but one that turned out to be useful) 06:29 ais523: OK 06:29 I also wanted to ask you something, regarding code quality 06:29 So, I want curses to be able to handle resizing without segfaulting 06:30 My current approach to this was to make a curses_getch() function which accepts a callback with a void * pointer 06:30 the idea being that if a KEY_RESIZE is called, it calls this callback, which varies depending on what dialog it is about 06:31 since different dialogs want to handle resizes differently 06:31 This means I'd have to create seperate structs for each kind of dialogs with the data they need to care about when resizing 06:32 Does this seem sensible? Or would it make more sense to handle KEY_RESIZE seperately for each kind of dialog? 06:37 here's an example of such a callback: http://sprunge.us/DGTS 06:48 ais523: sent devteam mail 06:48 FIQ: I have grunthack building with libuncursed using the regular makefiles. 06:48 It looks good... 06:49 Tangles: nice :) 06:49 "..." but? 06:49 Well the Makefiles don't look good. they are ugly as a camel's arse... but still. 06:50 When I play it, I can't dismiss popups with ESC anymore. 06:51 FIQ: looking now 06:51 Tangles: Hmm 06:51 What popups are you thinking of? 06:51 Because it might be because I messed with some/several of them and messed something up 06:51 inventory, or the options dialog... any of them. 06:52 Hmm 06:52 FIQ: how are you planning to handle the parts of the screen behind the displayed dialog? what if the dialog is in front of another dialog? 06:52 (or is it only ever possible to have one dialog up at a time? that would make things considerably simpler; it isn't true for nitrohack but I may take that approach with windowports I write in the future) 06:52 ais523: the only case where a curses dialog is above another one is during a listmenu + search dialog 06:52 so I figurd I'd handle that case specially 06:53 ah, OK 06:53 that's probably going to save yourself a lot of trouble 06:53 also I looked into how NH4 does it just now 06:53 in that case, I can see two approaches: a) a callback; b) recording the current state of the dialog, then just closing and reopening it 06:53 the approach seem to be *mostly* similar 06:53 the major difference is that NH4 will handle any kind of resize event 06:53 for *all* windows 06:54 by iterating a gw list 06:54 maybe I should do that too 06:54 that way I don't have to special case menulist+getlin 06:55 yep, NH4 basically has a window stack and will do redraws by iterating them all from bottom to top, a resize is similar 06:57 Tangles: hmm 06:58 Tangles: OK, can you try getlin in particular? I rewrote that from scratch earlier 06:58 maybe the old curses code for menulists use key events that work in curses but not libuncursed? 06:58 I'll check what curses is doing 06:59 FIQ: they probably use a hardcoded 27 06:59 KEY_ESCAPE is not 27 in libuncursed but probably is in curses 06:59 hmm but this is weird 06:59 I just looked 06:59 it seems to use KEY_ESC 06:59 and I can escape out of things just fine 06:59 KEY_ESC is not a valid spelling 06:59 but it uses KEY_ESC...? 06:59 hmm 07:00 ah 07:00 found it 07:00 in case it helps you debug: KEY_ESCAPE in libuncursed is 795, that's a number unlikely to be used by accident 07:00 include/wincurs.h 07:00 22:#define KEY_ESC 0x1b 07:00 seems like curses made its own KEY_ESC 07:00 ugh, that's painful and ridiculous 07:00 I'll fix that 07:01 looks like curses/ncurses doesn't have a KEY_ESC or KEY_ESCAPE at all 07:01 which is why people are hardcoding the 27 07:01 Tangles: Fixed. Try the last commit 07:02 (note: the least reliably readable keys on the entire keyboard, other than keys that you wouldn't expect to be able to read at all, are Escape and PrtSc) 07:02 hmm 07:02 but escape is kinda important 07:02 how does uncursed do it then? 07:03 timing 07:03 ah, it tries to figure out if the user is doing alt+stuff or escape? 07:03 it waits a few hundred milliseconds after seeing something that could be an Esc input, before it returns it 07:03 right 07:03 this is good enough for almost everyone, but casmith789 complained that he managed to "outrun" the Esc detector 07:03 so I had to add alt_is_esc 07:03 heh 07:04 I mean 07:04 I can also do it 07:04 but only if I actually try 07:04 ais523: maybe it is set a bit high? 07:04 -!- ais523 has quit [Quit: sorry for my connection] 07:05 -!- ais523 has joined #hardfought 07:05 ais523: maybe it is set a bit high? 07:05 lag don't usually slow down series of things 07:05 I can't remember the exact setting offhand 07:05 ok 07:05 but yeah, I can do it too, but I don't really consider it a problem 07:06 because I can only do it when I try to :P 07:06 FIQ: Where is the latest commit? 07:06 although lag can put a big cap in the middle if a packet break happens to happen there 07:06 that'd probably only be an issue with copy-paste though 07:06 *a big gap 07:06 Tangles: oh, my push was rejected 07:06 one moment 07:06 there 07:08 Tangles: I assume you are aware of the fact that your commit added a lot of trailing spaces in the makefile 07:08 Ah ok. it works. but there is a _noticeable_ delay. 07:08 Just a heads up 07:08 hmm 07:08 you're right 07:08 FIQ: I was not aware of the trailing spaces, sorry. 07:08 when ais523 said timing, I assumed like 4ms 07:08 or something like that 07:09 it's more like 400 07:09 almost half a second 07:09 seems to be around 250ms 07:09 ais523: this seems much too long? 07:10 I guess what we need is a test to see how quickly the individual characters of a complex keycode tend to arrive in practice 07:10 well it's definitely not *that* slow 07:10 IIRC curses uses 1000ms? might be wrong on that though 07:10 I'd imagine it to be nearly instantaneous 07:10 but having something like, say, 50-100ms just in case 07:10 but not 250 07:11 if you're on a 2400 baud modem then you're sending 10 or 11 bits per character and 2400 bits per second, so we're talking about 4-5 ms 07:11 I didn't really realize this because I tend to play on servers, which have innate lag anyway :P 07:11 but yeah, it's definitely a noticeable delay locally 07:11 so maybe the single-digit ms range is what we want 07:12 let me try 10ms and then mess around with test_uncursed over ssh on a bad connection 07:12 ok 07:12 that seems a bit on the short side, but if it works, it's all good 07:14 well it works locally, at least, I tested that first 07:15 Tangles: maybe the aimake-related things should all be moved into dumbmake.h 07:15 meh, it's all good I suppose 07:15 and AIMAKE_BUILDOS_linux should almost certainly *not* be in that file 07:16 Yeah. We can tidy it up. I just wanted to get it working first. 07:16 ok 07:16 still seems to work 07:16 let's commit this and see if anyone complains :-P 07:17 nice :) 07:17 ais523: do you distribute libuncursed seperately anywhere? 07:18 I ended up just copy/pasting over from fiqhack, but if a non-nethack project uses it, it shouldn't really need to pull nethack4 just to get libuncursed 07:18 FIQ: libuncursed1, no, it seemed simplest to have only one repository 07:18 ok 07:18 for aimake there's a separate repo and it's a pain keeping them in sync 07:18 git modules? 07:19 those are a pain purely by existing 07:19 hmm, true 07:19 and not only for yourself, for the downstream too 07:19 I remember working with a project that used them 07:19 it was really confusing 07:19 and more than once did I accidentally undo a module commit simply by not thinking of them 07:20 it seems like a good idea but they could really use less awkward handling 07:24 Tangles: cherry-picked ais523's last uncursed commit from nh4ais 07:24 try now 07:26 [nd] Eleven (Bar Hum Fem Neu), 27658 points, T:20417, killed by a touch of death 07:27 ais523: and in this case, cherry-picking is the only sane way to merge it :P 07:27 FIQ: seriously? it's at the tip of master 07:27 so how is a cherry-pick different form a regular merge? 07:27 ais523: Into GruntHack? 07:27 or are there older NH4 commits you haven't taken? 07:27 oh, into grunthack 07:28 yeah, lots of older NH4 commits that aren't in that :-P 07:28 for fiqhack I merged, yes 07:29 can you "merge" a single commit? 07:29 no 07:29 -!- elenmirie has joined #hardfought 07:29 -!- mode/#hardfought [+v elenmirie] by ChanServ 07:29 not in git 07:30 I guess the git way is to merge, then revert the ones you don't like 07:30 something like that 07:30 if you have an upstream that you got beef with 07:30 not in git 07:30 there are VCSes that can manage it 07:31 darcs seem cool 07:31 shpingrlde: no, the git way is basically just to hope 07:31 but apparently its UI kinda sucks 07:31 darcs has problems of its own, but it avoids most of git's 07:31 it's probably in your package manager, you could try it out if you like? 07:31 I intend to try it out at some point 07:31 I normally use darcs for personal projects 07:31 and git for projects that I expect other people might want to contribute to 07:31 or that might be widely used 07:32 apparently older versions of darcs had serious issues with bigger projects 07:32 not being able to scale, perf-wise 07:32 but it has gotten better 07:32 my main issue with darcs however seem to be its very dubious UI, from what ais523 told me 07:33 it's not dubious, it's actually quite nice (and some of git's best UI innovations seem to have been stolen from it) 07:33 it's just really bad at some very apparently basic tasks, which nonethless are rarely used in practice 07:33 ais523: I am talking about the way it asks you about merging things 07:34 Which seems insane 07:34 from what you told 07:34 git's UI for the same operation is much worse 07:34 the reason darcs' UI is complex is that the operation is complex 07:34 but if you just want to merge everthing you can just press a 07:34 it's only if you want to do a subset merge that things get mess 07:34 *messy 07:35 ais523: The UI isn't bad for what it is supposed to do @git 07:35 just that actually making a good UI for it becomes far harder for darcs 07:35 since it works completely differently 07:35 FIQ: looks good - the delay is noticeably shorter ;) 07:35 Tangles: apparently it was reduced from 200ms (I was off by 50) to 10ms 07:35 FIQ: right, what we have here is an operation that a) darcs can do but with an awkward UI, b) git can't do without a huge amount of manual work 07:36 ais523: git can't do it *at all* 07:36 you can rebuild the entire repository to put commits into the right order :-P 07:36 but yes, that's pretty close to "at all" 07:36 that... meh :P 07:36 incidentally, did I tell you I'm planning to write a program that rebuilds the entire repository to put commits in the right order? :-P 07:36 ais523: also, your suggestion for rewriting history in #nethack4 earlier 07:36 not sure if I ever responded properly to that 07:36 (yes I did) 07:37 but in the end I concluded that it was a bad idea 07:37 fair enough; I'd like to hear your counterarguments 07:37 because it is temporary *at best* 07:37 it would just resolve into the same thing again over time 07:37 hmm, so what if I wrote a git wrapper that automatically did the "rewrite" on commit (it wouldn't even be a rewrite in that case, because it'd be adding it in the right place the first time) 07:37 and it means that everyone who has a local 4k and FH repo needs to recreate theirs too from scratch 07:38 I think I agree that batch rewrites are a bad idea 07:38 ais523: that reminds me of something I was pondering myself 07:38 having a darcs-like handler as a "wrapper" 07:38 FIQ: I've already started writing the specs :-D 07:38 to git repositories 07:38 Heh :P 07:39 ais523: the main issue then becomes for people who don't want to bother with that 07:39 and might want to merge from FH/4k/NH4 07:39 I guess they'll just have to deal with cherrypicks 07:39 it's not like it could be handled any differently as-is atm 07:40 how big is our downstream, I wonder? 07:40 ais523: but for that I was rather thinking having *both* darcs and git repositories at the same time 07:40 rather than a special program 07:40 I did that for a while with acehack 07:40 and basically convert a darcs repo to a git repo 07:40 for the purpose of github hosting or similar 07:40 but the program I used to keep them in sync is unmaintained and now isn't in any of the repos 07:41 ahh 07:41 also, I suspect it only worked because acehack's history is really linear 07:41 it's literally linear, I just checked 07:41 at least after the conversion, the whole thing is one straight line 07:47 aosdict: Maybe make the formula have three components: (base cost for the spell) plus (failure rate * some constant) plus (failure rate * max Pw * some constant) ? 07:47 [nd] Terumasa (Val Hum Fem Law) killed Medusa, on T:41268 07:47 That'd let you tune that a little more finely. 07:58 -!- tacco\unfoog has joined #hardfought 08:08 [nd] Eleven (Pri Hum Fem Neu) rejected atheism by offering a gnome corpse on an altar of The Lady, on T:3367 08:11 shpingrlde: "merge" means "take all the commits that are in this branch that you don't already have", by definition. Taking just a single commit is cherry-picking. 08:15 -!- aoei has quit [Quit: Leaving] 08:19 -!- StinkyGreene has joined #hardfought 08:29 jonadab: using constants strikes me as something that won't scale very well, though I guess spell level factors into fail rate 08:29 hm, and I suppose with the wielding-spellbooks change, I did kind of backtrack the problem of it being too easy to sit around and farm utility spells 08:30 -!- ais523 has quit [Ping timeout: 240 seconds] 08:31 [nd] firemonkey (Wiz Elf Mal Cha) completed Sokoban, on T:9967 08:41 !tell K2 pushed a fix to the slex-2.1.1 branch that squashes a long-standing segfault when encountering a bones file on which the previous player died with a lit lightsaber in hand. 08:41 Will do, LarienTelrunya! 08:43 aosdict: The constant is just a scaling factor. 08:43 aosdict: And note that by * some constant I really meant * some constant / some constant 08:44 (Which is only even different because NetHack uses only integer arithmetic.) 08:47 [nd] Terumasa (Val Hum Fem Law), 233687 points, T:44007, quit 08:51 well, I'd like FIQ to weigh in since it was his idea originally 08:52 ? 08:58 23:53 FIQ: implemented your idea about making spells unfailable and increasing the Pw cost pro 08:58 portional to their failure rate... however, I'm not sure it works out in the original formulation 09:00 the problem with this is that low level characters have a rough time with even 2nd level spells 09:12 aosdict: interesting 09:13 aosdict: btw, what does 100% fail give? just "--Pw"? 09:13 anyway 09:14 the current NH3 spellcast failure calculation is very overcomplicated and doesn't really work all that well 09:14 I'd mess with that first 09:15 er 09:15 -!- raisse has quit [Ping timeout: 255 seconds] 09:15 -first 09:15 I think the idea itself is sound 09:15 just that it doesn't work with the old spellcast failure rates 09:16 And yes, the vanilla failure rate formula is pretty bad. 09:17 It's easily possible to be in a situation where a particular spell has a 98% failure rate, but increasing your intelligence by a couple of points _and_ gaining a couple of experience levels still leaves you with a failure rate in the high 90s. 09:17 And you basically never see failure rates in the vicinity of 50%. 09:17 Which is weird. 09:18 (The reason you don't is because of the hard spell / easy spell dichotomy, and the fact that some of the same factors used to calculate that are also used to adjust the failure rate subsequently. It's weird.) 09:20 hard spell / easy spell? 09:20 you mean where it decides whether a spell is "difficult" in percent_success? 09:21 FIQ: 100% fail gives infinity Pw cost, so it shows up on the menu as Inf. 09:22 aosdict: There's a threshhold, where if "difficulty" is higher than a certain amount, it uses a different formula. 09:22 jonadab: I seem to be pretty terrible at coming up with well-balanced formulas though, it'd really just be me guessing 09:23 It's not you. Coming up with well balanced formulas is inherently difficult. 09:23 I usually need 2-3 tries to get one that works. 09:23 And the jury's still out on some of mine, for that matter. 09:24 I guess the best technique is make a formula, create a bunch of scenarios for it, and see which of them are broken one way or the other. 09:24 Yeah. 09:24 Or create a formula with variables in it, and make a spreadsheet that plugs in different numbers for the variables. 09:24 And then tweak the formula until everything looks reasonable. 09:25 Actually, hmm. 09:27 Would you consider the part where it has different shield and metallic suit penalties for every role to be overcomplicated, or a necessary component of balance? 09:27 Not sure. 09:27 Let me play with this in a spreadsheet for a few minutes. 09:28 I think the Class Overhaul proposal had some ideas on how to simplify too. 09:30 Right, it contains: every spellcasting stat is Int, remove special spells, remove emergency spells, armor penalties are the same across all roles. 09:31 So role would only have an impact through the Base spellcasting penalty then? 09:31 That is worth considering 09:32 Actually, dtsund doesn't make any mention of that either. 09:40 need to go for a bit but I'm also setting up a spreadsheet to test various scenarios 09:52 -!- raisse has joined #hardfought 09:52 -!- mode/#hardfought [+v raisse] by ChanServ 09:54 -!- rikersan has joined #hardfought 10:11 izchak has 3 candles, 6 lamps 10:11 every single damn lamp is oil 10:11 -!- noty has joined #hardfought 10:37 Hmm, is there a list of median starting attribute values for each role? (Mainly interested in Int at the moment.) 10:40 jonadab: did an analysis of Dex once, might be able to work the same way with Int 10:42 yeah I think if you just looked up the values on the wiki and replaced them in this spreadsheet it would work 10:42 aosdict: ah 10:42 Though I guess it doesn't directly give the median. 10:42 FIQ: ? 10:43 the inf thing 10:43 jonadab: *cough*pw regen 10:45 FIQ: Pw regen will go off wis. 10:46 failure rate off int. 10:46 And yes, that means some roles may need adjustments to their attribute figures. 10:46 jonadab: so like FIQHack? 10:46 it uses wis and level there 10:54 Man, this stuff almost makes me wish for a six-dimensional spreadsheet. Except the my brain would explode. 10:57 oh no 10:57 https://www.reddit.com/r/roguelikedev/comments/7gz8m7/sharing_saturday_183/?st=japiu031&sh=393d9fa0 10:57 is linking to the evil wiki 10:57 bad aaron_ds, bad bad 11:17 aosdict: Perhaps something along the lines of percent_success_rate = int * 10 - penalty * 3 - splev * 25 11:17 what is "penalty"? 11:18 Where penalty depends on role, of course, and ranges from somewhere around -18 (monk with a robe and no armor) up to about 40 (monk wearing all the bad things and no robe) 11:18 The exact formula for penalty is negotiable; the player has significant control over it, based on what they wear. 11:18 Though there's a base number for each role that can't be dodged. 11:19 Which ranges from 1 (wiz) to 14 (bar) in vanilla. 11:19 spell level times 25? so even a high-level wizard with Int 20 and penalty 0 would have only a 25% chance of success? 11:19 aosdict: This is at unskilled. 11:19 Skill would improve it from there, haven't done a formula for that. 11:20 this means that level 7 spells have a 5% success rate at 18 int 11:20 with a penalty of 0 11:20 a level 1 spell has a 180% success 11:21 My numbers show Bar, Cav as the only ones completely unable to cast level 1 spells in the early game (given enough Pw; some other roles mail not have enough Pw at first). 11:21 worst-case monk but with 18 int can cast at 100%, 90%, 65%, 40%, 15%, 0%, 0% 11:21 And Wiz, Monk are the only ones who can potentially cast level 7 spells at unskilled in the late game, and that not at a favorable success rate. 11:22 FIQ: these are apparently unskilled numbers 11:22 And yes, Monks are significantly better casters than Wiz, it's that huge robe bonus; this could be tweaked. 11:22 jonadab: I said *worst-case monk* 11:22 butu with 18 int 11:22 Ah, right. 11:22 *but 11:22 it is not hard to max int in midgame with GA 11:23 oh I see the issue 11:23 you said penalty*3 11:23 But hmm, if we assume unskilled is 0, basic 1, skilled 2, expert 3, maybe divide the spell-level penalty by (1 + that) ? 11:23 ok, let's reset the numbers with this into account 11:24 then it becomes 35%, 10%, 0%, 0%, 0%, 0%, 0% 11:24 And yes, I'm playing with numbers here, the multipliers could maybe be tweaked. 11:24 seems more reasonable 11:24 Also, I've got experience level currently not doing anything, that could also change. 11:24 What I should do is write a script that creates a table given some numbers. 11:24 jonadab: I generally don't really consider attributes being supposed to have a large impact 11:24 due to how easy they are to max 11:25 Ah. 11:25 that's why wis doesn't do much for pw regen in fiqhack 11:25 So perhaps some of the impact that Int has here should go to XL? 11:25 (0.03*wis if more than 3) 11:25 -!- rikersan has quit [Quit: Lost terminal] 11:25 Eh, I need to go do some other things, but I will hopefully be back on late this afternoon. 11:28 I'm back now 11:28 and apparently jonadab just left 11:28 oh well 11:28 if only to-hit mattered, the Dex formula is kinda good for that "attributes shouldn't have a large impact" thing, FIQ 11:29 you get a lot of value out of > 18, which actually has a ring/gloves cost 11:31 dex is good for spellcasting and monks that are using body armor 11:31 beyond that, dex is pretty negligible beyond the early game 11:31 the other problem with dex is that its variance is extremely high among starting roles 11:32 so balancing to-hit that depends on dex is quite hard since Knights usually start with 7-9 and Rogues almost always start with 18 11:32 http://sprunge.us/bNjB 11:32 http://sprunge.us/ZLTR typo fix 11:32 reason why dex is mostly irrelevant later game with exception of spells and monk is because your XP level/Monster Dice give you a to-hit bonus equal to itself 11:33 For spells, you don't get this bonus 11:33 However, melee and ranged get the XP level/HD to-hit bonus 11:33 https://nethackwiki.com/wiki/To-hit 11:33 FIQ: I'd change the shield rules into "is wearing a shield" and separate penalty for "is wearing a metal shield" 11:33 making shields worse? D: 11:33 aosdict: elven shields aren't made of metal 11:34 in 4k, you could also shift some of the shield penalty into (lack of) shield skill 11:34 Webmant: I know, I'm trying to propose a whole new combat system, but to-hit is tough to figure out 11:34 Yeah, to-hit is strange 11:34 FIQ: yes, that's the point 11:34 very strange 11:34 differentiating, say, priests with good shield skill from wizards who generally want a free hand 11:34 only time i ever notice to-hit is with spells, very early game (once you level up a few times you start hitting very consistently), and with monks 11:35