00:50:49 -!- GiantOwl is now known as Kalir
00:53:16 <alexjurkiewicz> god "your mom"
01:24:42 <Gretell> Unstable branch on crawl.develz.org updated to: 0.22-a0-441-gcf5b0c6 (34)
01:57:23 <Gretell> Windows builds of master branch on crawl.develz.org updated to: 0.22-a0-441-gcf5b0c6
02:52:49 <Gretell> Monster database of master branch on crawl.develz.org updated to: 0.22-a0-441-gcf5b0c6
03:11:44 <Rotatell> Unstable branch on crawl.beRotato.org updated to: 0.22-a0-441-gcf5b0c6 (34)
03:53:30 -!- amalloy is now known as amalloy_
05:29:43 -!- Tux[Qyou] is now known as xXaMaNdA96Xx
05:30:05 -!- xXaMaNdA96Xx is now known as Tux[Qyou]
06:16:33 <Cheibriados> New branch created: pull/770 (5 commits) 13https://github.com/crawl/crawl/pull/770
06:16:33 <Cheibriados> 03Ryan02 07https://github.com/crawl/crawl/pull/770 * 0.22-a0-306-g5feaeea: First pass at a Sif Muna rework. 10(8 weeks ago, 11 files, 121+ 84-) 13https://github.com/crawl/crawl/commit/5feaeea47938
06:16:33 <Cheibriados> 03Ryan02 07https://github.com/crawl/crawl/pull/770 * 0.22-a0-307-g8f69cf9: Sif - slowed gift timer and slightly tightened Divine Focus cast timeout 10(7 weeks ago, 4 files, 3+ 3-) 13https://github.com/crawl/crawl/commit/8f69cf903f16
06:16:33 <Cheibriados> 03Ryan02 07https://github.com/crawl/crawl/pull/770 * 0.22-a0-444-gc158f9a: Merge branch 'master' of https://github.com/crawl/crawl into SifMuna 10(3 hours ago, 0 files, 0+ 0-) 13https://github.com/crawl/crawl/commit/c158f9a30927
06:16:33 <Cheibriados> 03Ryan02 07https://github.com/crawl/crawl/pull/770 * 0.22-a0-445-g4acf105: SifMuna: Balance tuning and code cleanup pass 10(28 minutes ago, 7 files, 31+ 77-) 13https://github.com/crawl/crawl/commit/4acf105cae0a
06:16:33 <Cheibriados> 03Ryan02 07https://github.com/crawl/crawl/pull/770 * 0.22-a0-446-g3093b7e: SifMuna - fixed a minor bug I had introduced 10(19 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/3093b7ecb00f
08:15:54 <Cheibriados> 03ebering02 07* 0.22-a0-423-gc9969fe: A knot of arrival vaults. 10(9 days ago, 1 file, 56+ 0-) 13https://github.com/crawl/crawl/commit/c9969fe3731e
08:15:54 <Cheibriados> 03gammafunk02 {GitHub} 07* 0.22-a0-443-g17c3723: Merge pull request #762 from ebering/arrival-vault-pr 10(9 seconds ago, 0 files, 0+ 0-) 13https://github.com/crawl/crawl/commit/17c3723931da
09:09:58 <Jorgrell> Unstable branch on crawl.jorgrun.rocks updated to: 0.22-a0-443-g17c3723931 (34)
09:29:30 <Cheibriados> New branch created: pull/771 (1 commit) 13https://github.com/crawl/crawl/pull/771
09:29:30 <Cheibriados> 03Muffindrake02 07https://github.com/crawl/crawl/pull/771 * 0.22-a0-442-g8c50517: refactor attack punctuation 10(11 minutes ago, 1 file, 4+ 10-) 13https://github.com/crawl/crawl/commit/8c50517b33ad
09:29:40 <advil> hm
09:32:24 <advil> I'm not sure it matters a lot there, but that's bog standard avoiding-float C code, not a "splinter of hot garbage"
09:33:06 <muffindrake> Psh
09:37:19 <muffindrake> The era of floating point being more expensive than your house is long over.
09:43:20 <muffindrake> advil: Actually, that isn't even C code
09:44:04 <muffindrake> No self-respecting C programmer would generate a string this way by appanding it multiple times
09:44:23 <advil> well, it's not c code in that string isn't c
09:44:59 <muffindrake> I'm talking about the general idea what the code is doing, but fine
09:45:29 <advil> the use of iterated >> is what I'm talking about, to be specific
09:47:07 <advil> I don't really mean to suggest that the existing code is a marvel of perfection
09:48:31 <muffindrake> Sure :p
10:00:37 -!- Guest13947 is now known as Lynn
10:01:07 -!- Lynn is now known as Guest31035
10:46:27 -!- Vaevictu1 is now known as Vaevictus
11:32:07 <muffindrake> minor mistake, oh well.
11:32:38 <muffindrake> I was too endazzled by 2 * HIT_STRONG
11:35:32 <Cheibriados> 03Muffindrake02 07https://github.com/crawl/crawl/pull/771 * 0.22-a0-443-g103f154: !!! 10(5 minutes ago, 1 file, 1+ 3-) 13https://github.com/crawl/crawl/commit/103f154cc27f
11:39:31 <muffindrake> Oh wait, HIT_MED is half of HIT_STRONG ...
11:46:02 <Cheibriados> 03Muffindrake02 07https://github.com/crawl/crawl/pull/771 * 0.22-a0-444-g83e4002: cut even more useless lines 10(5 minutes ago, 1 file, 1+ 3-) 13https://github.com/crawl/crawl/commit/83e4002e1edf
11:46:30 <muffindrake> If there is any way to reduce that function that I missed, tell me
11:47:33 <Pinkbeast> Doesn't this assume the relative values of HIT_MED and HIT_STRONG can't change?
11:47:56 <muffindrake> I don't think anybody _wants_ to change those macros
11:48:08 <muffindrake> They're used over the code base
11:49:25 <Pinkbeast> No, they aren't.
11:50:05 <muffindrake> 12 uses of HIT_STRONG with my changes
11:50:22 <muffindrake> 12 uses of HIT_MED
11:50:46 <muffindrake> and 8 HIT_WEAK
11:50:47 <muffindrake> Hmm
11:51:00 <Pinkbeast> They're used in 3 source files; in two, for production of exclamation marks; in one, for production of verbs ("impale"), etc.
11:51:20 <|amethyst> oh, hey
11:51:28 <Pinkbeast> That's not "all over the place" and if HIT_STRONG was changed one would expect all those things to change in concert.
11:51:42 <|amethyst> this thing and the one in constriction_damage_defender should probably be extracted into a function
11:51:54 <|amethyst> (in actor.cc)
11:52:44 <muffindrake> I just kinda assumed that those were macros whose use is fundamental to the code base
11:53:02 <muffindrake> but now I see that it is not and the code in the punctuation function is even duplicated in actor.cc
11:53:07 <|amethyst> hm, is there any reason for attack_strength_punctuation to be a method even?
11:53:07 <muffindrake> I'll get on that.
11:53:28 <|amethyst> I think it could be made static, or free-standing, without any problem
11:53:38 <|amethyst> then constriction_damage_defender can just call that
11:53:39 <Pinkbeast> ... or left alone since it works
11:55:00 <|amethyst> I'm not sure about using ilogb though
11:55:21 <|amethyst> sure, it's log base 2 everywhere we happen to build crawl right now
11:55:26 <|amethyst> but that's not guaranteed
11:55:45 <muffindrake> If you can point out an architecture that uses anything but FLT_RADIX == 2 and could run crawl, be my guest
11:56:42 -!- eb__ is now known as eb
12:00:12 <Pinkbeast> S/390.
12:01:24 <muffindrake> >Release date  September 5, 1990
12:01:50 <Pinkbeast> It's got a silly new name now but it's still about.
12:02:16 <|amethyst> Linux + gcc on S/390 does use base 2 though
12:02:50 <|amethyst> IIRC S/390 supports IEC 60559 (binary floating point) in addition to 360-style base-16
12:03:12 <|amethyst> but
12:03:17 <|amethyst> it's the principle of the thing :)
12:03:22 <Pinkbeast> |amethyst: Sure, but you might not want to run it on Linux. (I admit this is pretty unlikely, but it's a real extant architecture which could run Crawl, which kind of raises the question of why change something that a) works and b) is perfectly idiomatic?)
12:03:25 <|amethyst> there's already a log2 function
12:03:52 <muffindrake> Sure, I can use that instead. It's no big deal
12:03:58 <|amethyst> there are good reasons to change the fact that this same block of code occurs in two places
12:04:34 <|amethyst> since if someone changes the damage reporting scale, that would be two things that have to be modified
12:04:35 <Pinkbeast> |amethyst: That I don't disagree with (the instances in melee-attack.cc could also use a "count exclamation marks" function instead...)
12:04:54 <Pinkbeast> But just continuing to do it with iterated shifts and not eating HIT_STRONG might be better.
12:05:29 <muffindrake> perhaps keeping around HIT_STRONG is a better idea
12:06:00 <|amethyst> I'm not a huge fan of using floating point where it's not needed, but
12:06:21 <|amethyst> I bet these aren't the only two places that use iterated shifting to calculate an integer log base 2
12:06:49 <|amethyst> so at the very least *that* could be abstracted into a function
12:07:26 <muffindrake> I'll have that shortly
12:15:02 <muffindrake> Done.
12:16:16 <advil> some day I really need to merge my fixed point patch
12:17:32 <Cheibriados> 03Muffindrake02 07https://github.com/crawl/crawl/pull/771 * 0.22-a0-445-gba07374: change to free function and axe duplicate code 10(6 minutes ago, 4 files, 8+ 18-) 13https://github.com/crawl/crawl/commit/ba073746f924
12:17:32 <Cheibriados> 03Muffindrake02 07https://github.com/crawl/crawl/pull/771 * 0.22-a0-446-g597e5f9: axe include already included by ranged-att header 10(4 minutes ago, 1 file, 0+ 1-) 13https://github.com/crawl/crawl/commit/597e5f9ea62c
12:19:17 <muffindrake> advil: What is that about?
12:39:23 <muffindrake> #define XOM_SPEECH(x) x
12:39:26 <muffindrake> ?
12:41:45 <|amethyst> %git 29d148a2
12:41:45 <Cheibriados> 07kilobyte02 * 0.14-a0-1789-g29d148a: Let db_lint check Xommy parts of godspeak.txt. 10(4 years, 5 months ago, 2 files, 35+ 18-) 13https://github.com/crawl/crawl/commit/29d148a2b9f3
12:43:02 <|amethyst> allows db_lint to identify those to make sure they actually appear in the database files (and, conversely, that there aren't unused speech types in the database files)
12:43:23 <muffindrake> Ah, I see
12:49:05 <Cheibriados> When I try to go upstairs, the game crashes 13https://crawl.develz.org/mantis/view.php?id=11486 by Siegurt
13:05:28 <Henzell> Unstable branch on crawl.akrasiac.org updated to: 0.22-a0-443-g17c3723 (34)
13:31:12 -!- amalloy_ is now known as amalloy
14:12:07 <advil> muffindrake: it's a revamp of the fixed point math that crawl does to be more systematic, would be a natural place to write a >>-based integer log function
14:16:02 <muffindrake> advil: Is that intended to improve clarity or simply make it consistant across the code base?
14:16:22 <advil> both
14:19:10 <muffindrake> What is it with me swapping e's with a's today
14:19:30 <advil> it gets rid of all the ad hoc scale arguments in the uglier parts of the code
14:20:02 <advil> here's an example https://github.com/rawlins/crawl/commit/eb32a295a3c1cdcd0a743cfdac471eb0df789156
14:22:18 <amalloy> |amethyst: if we're extracting a function for integer log2, what do you think of using __builtin_clz instead of doing repeated shifting? my understanding is that both gcc and clang include this builtin, and use hardware support when possible
14:24:22 <muffindrake> advil: That does look nice
14:26:40 <muffindrake> I'm a bit disturbed by the fact that crawl code uses int _everywhere_, even though int only has a small guaranteed range; and seemingly oblivious to the fact that integer overflow is undefined
14:27:50 <amalloy> you mean, as opposed to using uint32_t or something similar?
14:28:26 <muffindrake> Those might just fit better, yes
14:28:58 -!- Jozrael_ is now known as Jozrael
14:30:20 <amalloy> do you have specific cases in mind that seem oblivious to integer overflow?
14:31:55 <amalloy> i think we would be happy to fix as many of those as we can find, but i don't like the idea of switching to fixed-size representations
14:33:26 <advil> "save file corrupted -- empty block" anyone have guesses as to what would cause this?
14:34:10 <amalloy> advil: i posted a note to the bug a minute ago
14:34:16 <advil> ah ok
14:34:30 <amalloy> i learned some things about crawl's save file format this morning, but evidently not enough
14:34:40 <advil> heh
14:34:42 <advil> that is a spooky bug
14:34:52 <advil> I hope it's not our fault
14:34:57 <amalloy> well, we wrote the program
14:35:43 <amalloy> the weird thing to me is the save file has correct-looking metadata, like "D:11 starts at offset X, D:12 starts at offset Y", and so on
14:36:00 <amalloy> and those numbers look reasonable. but then when you load the chunk header that starts at offset X, it says "this chunk is 0 bytes long"
14:38:07 <advil> I mention "not our fault" because (a) if it's our fault it's probably my fault, realistically, and (b) the source comment next to that message is "// This reeks of on-disk corruption (zeroed data)."
14:38:32 <advil> well, I guess I don't know what version that save might originate from
14:39:23 <amalloy> hah, did you do something recently to save file marshaling or something?
14:39:53 <advil> yeah, but I did it just by adding a prop so it should have been safe
14:40:11 <advil> afaik that's the only recent change to save format though
14:43:19 <advil> I wonder why "corrupted" doesn't produce a crash dump
14:44:35 <amalloy> it probably gets called in cases where the stuff we normally include in crash dumps might be unavailable
14:44:42 <amalloy> so you'd want to produce a more minimal crashdump
14:44:59 <advil> ah, makes sense
14:45:56 <muffindrake> Well, crawl hasn't aborted yet with -ftrapv, so I'll have to search harder
14:48:37 <MarvinPA> !tell gammafunk i think i'd still tend towards removing jorgrun, since he was added just for shatter, and even with some hypothetical other damage spell replacing it for him there's a bunch of other similar uniques at that level already
14:48:38 <Sequell> MarvinPA: OK, I'll let gammafunk know.
14:49:58 <MarvinPA> !tell gammafunk but either way removing shatter as a monster spell fully (for ghosts too if they can get it) sounds right
14:49:59 <Sequell> MarvinPA: OK, I'll let gammafunk know.
14:55:17 <muffindrake> The last time I got jorgrun in my game, he killed himself with LRD, is that intended?
14:59:11 <MarvinPA> possibly, to the extent that monsters can decide to hit themselves with explosions if they think it'll sufficiently more damage to you/your allies than it will to them
14:59:18 <MarvinPA> who knows exactly how that actually works though
15:11:24 <geekosaur> it's a long running confusion to newish players that high level uniques and pan lords often kill themselves with spells aimed at the player
15:12:04 <Pinkbeast> Even when I was very new I dunno that I complained when it happened
15:14:21 <muffindrake> I wasn't in any danger from dying to jorgrun, as he would have blown up without getting a single move on me after a successful LRD
15:14:28 <muffindrake> merely confused
15:15:31 <amalloy> ??jorgrun[shatter
15:15:33 <Sequell> jorgrun[1/8]: A dwarven earth mage, with Shatter, LRD, petrify, and dig.
15:17:01 <amalloy> i was expecting something there showing his damage dice for shatter. imo most players are in danger of dying to jorgrun almost any turn he is in LOS, since you could get unlucky energy randomisation into double shatter
15:17:39 <Pinkbeast> I don't want to be too radical but couldn't Shatter just do less damage to player Gr ?
15:21:45 <MarvinPA> it would still be a bad design for a monster spell if that were the case
15:22:17 <MarvinPA> gargoyles aren't really the problem in any form
15:27:21 <geekosaur> earth spells in general are all kinda absurdly powerful (there's a reason they all get triple AC check among other things)
15:45:19 <amalloy> all? i was only aware of LRD and /scattershot getting triple AC checks
15:46:42 <amalloy> oh, apparently sandblast too
15:49:03 <amalloy> huh, and qazlal's earth-flavoured upheaval too. but not iron shot and such. certainly not shatter
15:50:50 <amalloy> weirdly, shatter skips the usual AC code entirely, instead rolling for monster AC manually and then directly reducing the monster's HP.
16:11:15 <MarvinPA> triple-ac is just for shrapnel-flavoured things yeah
17:06:23 <minmay> amalloy: energy randomization can't cause that
17:06:36 <amalloy> oh, really? is it just moves?
17:06:39 <minmay> yeah
17:06:51 <amalloy> TIL. that makes sense, i guess
17:07:23 <minmay> you could get double-spelled or double-meleed by orc priests, hydras, etc. as a human or whatever in 0.6, but that was changed fairly quickly in 0.7
17:07:58 <minmay> monsters can get move+spell/attack or double move but not double spell/attack, in layman's terms
18:20:43 <Lantell> Unstable branch on underhound.eu updated to: 0.22-a0-443-g17c3723931 (34)
19:01:34 -!- amalloy is now known as amalloy_
23:59:29 <Eksell> Unstable branch on CRAWL.XTAHUA.COM updated to: 0.22-a0-443-g17c3723 (34)