00:06:51 <switch__> not certain there's much left to do regarding the MSVC PR
00:07:18 <switch__> can probably review/pull it whenever someone gets a chance
01:04:27 <Eksell> Unstable branch on CRAWL.XTAHUA.COM updated to: 0.23-a0-386-g9c9d49f (34)
02:27:29 <Gretell> Unstable branch on crawl.develz.org updated to: 0.23-a0-386-g9c9d49f (34)
02:58:41 <Gretell> Windows builds of master branch on crawl.develz.org updated to: 0.23-a0-386-g9c9d49f
03:11:46 <Rotatell> Unstable branch on crawl.beRotato.org updated to: 0.23-a0-386-g9c9d49f (34)
03:59:09 <Gretell> Monster database of master branch on crawl.develz.org updated to: 0.23-a0-386-g9c9d49f
05:22:42 -!- amalloy is now known as amalloy_
06:30:11 <ttky123> hi
07:26:39 <ttky123> hi
07:42:08 <simbs38> hello there, is there anyone there?
07:42:39 <ttky123> I'm here
07:43:17 <simbs38> hi mate, i want to build a bot to play crawl, can u guide me please? i dont even know how to start
07:43:46 <ttky123> ugh sorry I'm here to learn how to participate to this game too
07:44:45 <simbs38> thanks anyway, good luck to you, anyway else here? xD
07:47:52 <simbs38> hi there cerpin
08:03:56 <simbs38> hello?
09:16:13 <alexjurkiewicz> ??qw
09:16:14 <Sequell> qw[1/7]: A fully automated lua bot written by elliptic, with some code borrowed from parabolic and xw. See "!lg qw won 2" for the first ever completely uninterrupted and unassisted bot win. Git repository: https://github.com/elliptic/qw
09:16:19 <alexjurkiewicz> simbs38: ^
09:22:16 <ttky123> still nothing?
09:48:44 <simbs38> hi there, i saw that bot, but that has 8600 lines of code, i would like to start with something more simple to learn the api and want i can and canot do
09:53:50 <ebering> if you have the source, the lua api is documented using ldoc
09:53:57 <ebering> if you have ldoc installed you can run 'make api'
09:54:05 <ebering> and it will compile the documentation in to the docs/lua folder
09:54:46 <ebering> a (not automatically updated) host for them is lua.dcss.xyz
10:12:59 <simbs38> that helps a lot, thanks, but how can i start my bot? is there any skeleton of code of a basic bot? and how to send comands to the game?
10:16:56 <ebering> simbs38: if you wnat to write a bot in lua, you put your lua code in the .crawlrc file
10:17:12 <ebering> crawl will call the function named ready() in that file when it wants input from you
10:17:37 <ebering> ??xv
10:17:37 <Sequell> xv[1/1]: "x-v", it means to examine a monster. Use it to see how quickly a ghost is going to kill you.
10:17:41 <ebering> rip
10:17:45 <ebering> wait
10:17:46 <ebering> ??xw
10:17:47 <Sequell> xw[1/3]: D->lair->D:14->(Snake or Spider)->Vaults:1-4->Orc:1-3->Depths->???
10:17:49 <ebering> ??xw[2
10:17:50 <Sequell> xw[2/3]: <N78291> ah, xw is trying to eat a wand of disintegration
10:17:52 <ebering> ??xw[3
10:17:52 <Sequell> xw[3/3]: <conted> It's like autoexplore <conted> But more advanced
10:17:59 <ebering> ??gw
10:17:59 <Sequell> gw[1/10]: http://crawl.berotato.org/crawl/rcfiles/crawl-0.17/gw.rc
10:18:04 <ebering> there's another bot
10:18:06 <ebering> different from qw
10:19:33 <simbs38> in the qw bot to run it we use ./crawl -rc qw.rc, can't i run my bot like that to?
10:19:45 <ebering> yes
10:20:47 <simbs38> the qw bot ask to pres TAB from time to time to run, why is that?
10:21:08 <simbs38> why doesn't it run automatic?
10:25:34 <ebering> I'm not superfamiliar with qw, but part of it is when it's running on an online server to ratelimit it (there's an Expect script that drives the ssh connection), also to allow debugging and observation
10:30:15 <simbs38> im running it offline
10:31:34 <ebering> well, then you can turn it off
10:32:25 <simbs38> how?
10:33:28 <ebering> by editing qw.rc
10:33:35 <ebering> like I said, I'm not super familiar with it, Icouldn't tell you what lines exactly to change
10:39:27 <simbs38> ok, thanks mate u gave me a good starting point
10:47:28 <Rotatell> mandarbmax (L20 MfGl) ERROR in 'mon-util.cc' at line 664: bogus mc (no monster data): invalid monster_type 1000 (1000) (Depths:2)
10:50:53 <ebering> hm. that's the second one of those
10:50:56 <ebering> !crashlog
10:50:58 <Sequell> 20356. mandarbmax, XL20 MfGl, T:41973 (milestone): http://crawl.berotato.org/crawl/morgue/mandarbmax/crash-mandarbmax-20181028-144718.txt
12:35:55 <ebering> A
12:56:13 <Cheibriados> 03stenella02 07https://github.com/crawl/crawl/pull/879 * 0.23-a0-387-gbf9d4db: Console support 10(16 hours ago, 8 files, 329+ 63-) 13https://github.com/crawl/crawl/commit/bf9d4dbfbdf5
13:05:40 <Henzell> Unstable branch on crawl.akrasiac.org updated to: 0.23-a0-386-g9c9d49f (34)
13:43:37 <Cheibriados> FR for offline versions: add a grand total score of top 100 games 13https://crawl.develz.org/mantis/view.php?id=11735 by RoGGa
15:08:15 <Cheibriados> Deep elf elementalist casts `Awaken Earth` through transparent stone wall 13https://crawl.develz.org/mantis/view.php?id=11736 by Petrovich
15:18:24 <MarvinPA> that's how it's meant to work, right? since it doesn't require los generally
15:18:42 <MarvinPA> (i'm all for fixing it by removing the spell though)
15:25:24 <advil> yes, but I can see the intuition that you ought to be able to flee from it by hiding behind enough stone; wonder what happens with non-transparent stone
15:41:44 <MarvinPA> i think it can just affect any tile in a line between you and the caster, regardless of visibility
15:44:56 <advil> killholes or no killholes, it does seem like a bad idea to me to have a spell like that because it's super dependent on hidden monster tracking stuff
15:45:32 <advil> i.e. monsters knowing locs of recently seen players out of los
15:46:34 <advil> well, I guess it's no different from dig, just smite targeted
15:46:47 <advil> so maybe it's the smite targeting + that property that bugs me
15:46:50 <advil> %git 3704a81a433fd
15:46:50 <Cheibriados> 07Grunt02 * 0.18-a0-777-g3704a81: Monster spell: Awaken Earth. 10(2 years, 11 months ago, 6 files, 169+ 3-) 13https://github.com/crawl/crawl/commit/3704a81a433f
15:47:25 <MarvinPA> i think it's just a weird approach to solving however much of a problem killholes are, yeah
15:47:53 <MarvinPA> i like making player digging substantially rarer (which i think gammafunk was planning on doing to the wand?)
15:48:53 <advil> yes, that seems pretty reasonable to me
15:56:45 <minmay> ok I feel like this problem is my fault: kingbuddyboy pointed out that you can now wield a +Rage weapon (like glaive of the guard), evoke it, and then swap to the weapon you actually want to use
15:58:15 <minmay> also, peanut gallery comment: i strongly suspect that deep elf elementalists would work better with just regular dig spell
16:26:49 <gammafunk> MarvinPA: here's a summary of something I sent you some time back about wand charge reductions
16:27:00 <gammafunk> I'm thinking of reducing dig charges to generate at the same level as scattershot per wand (5), disint/enslave/para/poly down to 8 per wand, and maybe flaming/random effects down to 16.
16:27:04 <gammafunk> The digging reduction is pretty aggressive, gets us down to made 25 charges for a 3-rune game on-average. For the others mentioned it's a 33% reduction.
16:27:07 <gammafunk> Wand charges feels pretty high still, but it's mostly the hex wands that feel perhaps too high.
16:27:10 <gammafunk> Give us maybe 32 charges in a game for e.g. enslavement on-average, and about 48 for paralysis.
16:28:04 <gammafunk> I can do just the digging adjustment
16:28:14 <gammafunk> But unless people have reservations about those other changes, I can do those as well
16:28:42 <gammafunk> and feel free to speak up if you think adjustments downward are ok but need to be done with some of the details differently
16:29:58 <gammafunk> wasn't planning to change the direct attack wands other than flaming. I guess disint is an attack wand, but MR-checking wands feel like they have too many charges in the game
16:30:15 <MarvinPA> those look pretty reasonable at a glance, those are the average values on generation right? rather than max charges
16:31:09 <MarvinPA> oh i'm out of date in how i'm thinking about them anyway right, i forgot about stacking
16:32:45 <MarvinPA> i guess i would maybe tend towards having disint get more charges than the other MR-checking wands? since it's not quite as overwhelmingly strong an effect
16:33:28 <MarvinPA> hmm displaying the old max charges on the \ screen is kinda meaningless now
16:42:45 <gammafunk> yeah, that's average I guess
16:42:58 <gammafunk> man, I made that stacking change but I forget how the charges are done; it's totally randomized right?
16:43:07 <gammafunk> not sure what the range is
16:43:43 <gammafunk> yeah, disint could remain the same
16:43:43 <gammafunk> let me see what they all are
16:48:05 <gammafunk> default is 24 charges, clouds/scattershot get 9, acid/iceblast get 15, flame/randomeffects get 48
16:48:25 <gammafunk> hrm
16:48:30 <gammafunk> so the average is half of that
16:48:35 <gammafunk> those are actually max charge values
16:49:00 <MarvinPA> yeah, then generation is fairly weighted towards the middle of the range up to that max
16:49:00 <gammafunk> it's 1 + random2avg(MAX, 3)
16:49:00 <gammafunk> right
16:49:19 <gammafunk> so we could move dig to the cloud/scattershot tier
16:49:46 <gammafunk> I guess enslave/para to the acid/iceblast tier
16:50:27 <gammafunk> I suppose poly as well
16:50:41 <gammafunk> then remove the 48 tier for flame/randomeffects
16:51:06 <gammafunk> er I guess my plan was to make that 32 instead of 48
16:51:27 <MarvinPA> that sounds good yeah, probably fine for them to still have a bit more than average
16:51:58 <gammafunk> alright, I'll go with that and if anyone wants to adjust further, they could
17:25:19 <switch__> this tukimas bug is nuts
17:29:08 <switch__> tukimas works properly when it selects the 2nd of 2 weapons. it reproduces the bug whenever it selects the 1st of 2 weapons. in theory it should always select the 1st of 2 weapons, but for some reason sometimes it does not
17:30:01 <switch__> forcing it to select the second weapon (if it exists) on monsters that carry two weapons works most of the time, but some sizable fraction of the time the remaining weapon is unassigned/corrupted
17:30:33 <switch__> o____o
17:33:47 <switch__> this is getting into the weirder areas of code that i have absolutely no experience in
17:40:43 <gammafunk> !clap crawlcode
17:40:43 <Sequell> 👏 c 👏 r 👏 a 👏 w 👏 l 👏 c 👏 o 👏 d 👏 e 👏
17:43:47 <switch__> .___________.
17:47:58 <Lantell> MeekVeins (L21 DrFE) ERROR in 'mon-util.cc' at line 664: bogus mc (no monster data): invalid monster_type 1000 (1000) (Vaults:5)
17:51:15 <gammafunk> !crash meekveins
17:51:18 <Sequell> 1. MeekVeins, XL21 DrFE, T:49164 (milestone): https://underhound.eu/crawl/morgue/MeekVeins/crash-MeekVeins-20181028-214756.txt
17:51:29 <gammafunk> doh
17:52:28 <gammafunk> hrm signal #6 aborted
17:52:34 <switch__> oh, i figured it out.
17:53:01 <switch__> it selects either the first or the second weapon based on an earlier call to target->weapon() which chooses between the first and second randomly
17:53:28 <switch__> so the broken case was when this call selected the first one, but i forced the second one
17:53:36 <switch__> h e h
17:54:00 <switch__> the problem is there are 25 other things that call weapon() and i don't want to fuck with that
17:56:42 <switch__> (the actual visual bug seems to be due to the way dual-wielding works in tile generation; the second weapon fits in the "shield" attribute, but when the first weapon is unequiped, the second one is moved to the first slot, so it's still the shield attribute that's nullified. the next time the tile is redrawn, it works correctly
17:57:45 <Henzell> BerryKnight (L18 HOAr) ERROR in 'mon-util.cc' at line 1568: bogus mc (no monster data): invalid monster_type 1000 (1000) (Orc:2)
17:58:37 <gammafunk> ah
17:58:42 <gammafunk> !crash berryknight
17:58:52 <Sequell> 1. BerryKnight, XL18 HOAr, T:29664 (milestone): http://crawl.akrasiac.org/rawdata/BerryKnight/crash-BerryKnight-20181028-215728.txt
17:59:01 <gammafunk> ok, so another crash
17:59:07 <gammafunk> probably something happening with these god conducts then
17:59:24 <gammafunk> hrm, this one might not be related
17:59:52 <gammafunk> maybe this is memory corruption
18:01:00 <gammafunk> monster soul eater using drain life
18:01:14 <gammafunk> will definitely take a closer look at that
18:02:59 <Cheibriados> New branch created: pull/881 (2 commits) 13https://github.com/crawl/crawl/pull/881
18:03:00 <Cheibriados> 03Skrybe02 {GitHub} 07https://github.com/crawl/crawl/pull/881 * 0.23-a0-387-gd7aed64: Address a TODO in layout_cave_pools 10(38 minutes ago, 1 file, 7+ 5-) 13https://github.com/crawl/crawl/commit/d7aed64223c7
18:03:00 <Cheibriados> 03Skrybe02 {GitHub} 07https://github.com/crawl/crawl/pull/881 * 0.23-a0-388-g61f66aa: Address a TODO in layout_concentric_octagons 10(32 minutes ago, 1 file, 5+ 6-) 13https://github.com/crawl/crawl/commit/61f66aafc122
18:11:22 <switch__> hmm
18:11:28 <switch__> what do you think is the correct way to fix this bug
18:13:40 <gammafunk> If you're talking about the drain life thing, I'll fix it after I push another commit
18:13:47 <switch__> no, tukimas
18:14:06 <gammafunk> oh, haven't read that code, sorry
18:14:18 <switch__> basically the bug is that the object that should be in the second hand is now in the first
18:14:24 <switch__> but the game doesn't know it yet so it doesn't draw it
18:27:37 <Lantell> MeekVeins (L23 DrFE) ERROR in 'mon-util.cc' at line 664: bogus mc (no monster data): invalid monster_type 1000 (1000) (Depths:3)
19:18:19 <alexjurkiewicz> any plans for the datafiles merge?
19:18:55 <alexjurkiewicz> ideally it can be merged at a time when I'm around, and at least an hour before every other server rebuilds, cos I can watch CPO's compilation and spot any immediate issues
19:21:10 <Lantell> Unstable branch on underhound.eu updated to: 0.23-a0-386-g9c9d49f623 (34)
19:57:14 <gammafunk> alexjurkiewicz: I think advil is still planning to give it another once-over as some point; we wanted to have at least one dev around to address issues
20:00:28 <ebering> gammafunk: do you have thoughts on 881? It looks good to me but I'm not super familiar with this part of the level builder
20:00:33 <ebering> and dunno if there are gotchas
20:00:43 <gammafunk> ah, yeah, I can take a look at that soonish
20:00:51 <gammafunk> I have a little bit of a backlog
20:01:12 <gammafunk> It's probably a fairly minor change
20:01:26 <gammafunk> perhaps a bit hard to test it, but you could just max weight on the affected layouts
20:01:30 <gammafunk> and generate some levels
20:01:32 <gammafunk> and look at them
20:01:44 <gammafunk> but that depends on the nature of the changes, which I haven't even looked at
20:02:02 <ebering> yeah I've done that
20:02:08 <ebering> uh, the mantis ticket has some screenshots
20:02:38 <alexjurkiewicz> I'm writing a patch for hellcrawl which modifies armour in the player's inventory, and it keeps duplicating the inventory item (so you have two items in inventory which are `b - a +0 scale mail`. Can anyone take a look at the code and tell me why? https://github.com/alexjurkiewicz/crawl-ref/blob/anvil/crawl-ref/source/shopping.cc#L1381
20:02:42 <ebering> https://crawl.develz.org/mantis/view.php?id=11214
20:03:07 <alexjurkiewicz> I think it's something about how I convert from `int item_slot` into `item_def& item`
20:03:32 <alexjurkiewicz> (which I do because I want to check the item's base_type and so on)
20:03:40 <alexjurkiewicz> (and then modify the item)
20:06:27 <gammafunk> the screenshots do look nice but yeah, I'd need to read the commit in detail to comment further
20:07:19 <geekosaur> alexjurkiewicz, does the first item in inventory also disappear?
20:07:50 <geekosaur> you an't assign to refs normally, I think 'item = ...' is overwriting u=you.inv[0]
20:08:07 <geekosaur> s/u=//
20:08:46 <geekosaur> may need to use a pointer instead of a ref, or declare and initialize item there instead of early
20:13:19 <ebering> I think I'm gonna run a mapstat run with 881 added in and if it runs well merge it
20:13:50 <ebering> I agree with the layout writer of yore who thought these would be good D/Lair layouts
20:21:41 <geekosaur> in fact I dont' just think (had to swap c++ brain back in), tht is what's happening once initilized, a ref is immutable and assignment modfiies the thing referenced, so it's writing over you.inv[0]
20:22:03 <geekosaur> s/happening once/happening. once/
20:25:53 <geekosaur> https://en.cppreference.com/w/cpp/language/reference_initialization
20:26:14 <geekosaur> (also the declaration should be "free", so there's no point in declaring it outside the loop)
20:32:28 <alexjurkiewicz> ah
20:32:37 <alexjurkiewicz> ok, let me try cleaning up the code
20:32:56 <alexjurkiewicz> is there some way in a loop like that to say "this variable is constant for each iteration of the loop"?
20:33:39 <geekosaur> references are alwasy constant
20:33:39 <alexjurkiewicz> I guess pull that code into a sub-function
20:33:45 <geekosaur> just declare it where you need it
20:34:11 <geekosaur> that's what I ment by "free"; there's no guarantee it actually represents anything in memory
20:35:00 <geekosaur> (this is also why "assigning" to iot overwrites what it's referencing; it's a alias, not a variable)
20:35:09 <geekosaur> an alias
20:36:05 <geekosaur> more genrally, you only get "constant per iteration" by declaring it inside the loop
20:36:59 <geekosaur> I would not be afraid of doing tis in general, if the compiler's any good it recognizes that it can allocate once and overwrite per loop iteration
20:37:05 <alexjurkiewicz> so I want to be able to check the item's base_type inside the loop, but then still have that variable be accessible once I exit the loop
20:37:45 <alexjurkiewicz> so I suppose I need an `item_def` variable outside the loop which I assign to, and I can &-ise it inside the loop temporarily
20:38:49 <geekosaur> think you just want a pointer here, not a reference
20:38:56 <geekosaur> initialize to NULL
20:39:16 <alexjurkiewicz> yeah, you are right
20:39:18 <geekosaur> then item = &you.item[item_slot]
20:39:41 <geekosaur> and if you want to avoid all the *s, initilize a reference from the pointer right after the loop
20:40:17 <geekosaur> so itemp = ..., then item_def &item = *itemp
20:46:43 <alexjurkiewicz> thanks! that looks better now
20:53:34 -!- neunon_ is now known as neunon
21:39:54 <switch__> fixing the bug is pretty simple, i can do it with 4 or 5 lines of code
21:39:54 <switch__> but it's super ugly
21:40:04 <switch__> need a dev's advice on 858
21:40:17 <switch__> or rather your blessing to make ugly changes
21:40:34 <switch__> or the call to rewrite the entire module
21:45:43 <ebering> I'll wager there's a solution in between the two
21:45:59 <ebering> cleaner than what your first idea is, but far short of a complete re-write
21:46:23 * geekosaur can think of a few source files where that stands a god chance of being false
21:46:31 <geekosaur> *good
21:48:05 <switch__> it feels like some of these bugs are just outside of the edge of the implementation
21:48:22 <switch__> like the person that put these bugs in should have seen that the codebase wasn't there for it, and rewrote it to compensate
21:49:36 <switch__> like, regardless of this bug right now, the way the files are written don't allow for more than 2 weapons
21:50:29 <ebering> that's a reasonable assumption...
21:50:29 <geekosaur> the original design only allowed for one ranged and one melee weapon
21:50:34 <geekosaur> (monsters)
21:51:03 <switch__> i mean, they're given 12 inventory slots
21:51:09 <switch__> it's a fantasy game
21:51:14 <switch__> maybe someone has 7 hands
21:51:16 <switch__> you don't know them
21:51:32 <switch__> oh shit, that's an idea, isn't it
21:51:53 <geekosaur> octopodes :p although dual wield is itself a hack.
21:52:18 <geekosaur> whch is the real problem, this is an ancient code base that has been perverted into doingthings it was never designed for
21:52:40 <switch__> it just seems like nobody along the way was like "maybe we should make this scalable"
21:53:33 <geekosaur> welcome to crawlcode
21:53:39 <switch__> stop saying that
21:53:44 <geekosaur> design has never been a thing to speak of
21:56:29 <ebering> ??linley[6
21:56:29 <Sequell> linley henzell[6/7]: Henzell discussing his programming ability: "You see, I gave up on Borland's [C++] manual (which is very good, I'm just lazy) when it started talking about pointers [....] After pointers came structs, enums, pragmas, classes, macros, and a whole lot of other stuff which still means nothing to me."
21:56:41 <ebering> the original author of Crawl
22:07:24 <gammafunk> got to learn those pragmas
22:38:41 <alexjurkiewicz> #pragma DWIM
22:55:07 <ebering> ok, mapstat finished. 881 seems good to go so I'm gonna smash that merge button
22:55:35 <Cheibriados> 03gammafunk02 07* 0.23-a0-387-g141017e: Lower the average number of charges for some wand types 10(5 hours ago, 1 file, 5+ 1-) 13https://github.com/crawl/crawl/commit/141017ece01d
22:56:08 <ebering> oh man that'll make some players mad
22:56:36 <gammafunk> well
22:57:00 <ebering> nerfs do. it's a necessary change
22:57:10 <gammafunk> we'll have to keep improving unrands
22:57:12 <ebering> no more wand of digging (77)
22:57:20 <gammafunk> players are easilly fooled by shiny things
22:57:28 <gammafunk> yeah, that could still happen of course, but definitely won't be the norm
22:57:37 <Cheibriados> 03Skrybe02 {ebering} 07* 0.23-a0-388-g786a77d: Address a TODO in layout_cave_pools 10(6 hours ago, 1 file, 7+ 5-) 13https://github.com/crawl/crawl/commit/786a77dc79ec
22:57:37 <Cheibriados> 03Skrybe02 {ebering} 07* 0.23-a0-389-g4a5eb33: Address a TODO in layout_concentric_octagons 10(5 hours ago, 1 file, 5+ 6-) 13https://github.com/crawl/crawl/commit/4a5eb334b0ed
23:10:49 <Jorgrell> Unstable branch on crawl.jorgrun.rocks updated to: 0.23-a0-389-g4a5eb334b0 (34)
23:23:58 <Cheibriados> vault generating with no way in (if only I could dig with lrd, GAMMAFUNK) 13https://crawl.develz.org/mantis/view.php?id=11737 by paterick
23:27:46 <doy> haha, looks like he's still playing https://questionablecontent.net/view.php?comic=3861
23:29:32 <ebering> indeed