00:36:58 rocky1 (L23 HOFi) ERROR: range check error (20 / 20) (D (Sprint)) 00:50:05 aidanh: could expose pause/resume rendering to lua 00:50:41 that has issues with lua spamming redraws but 01:35:20 Unstable branch on crawl.develz.org updated to: 0.26-a0-393-g67388ea477 (34) 02:04:47 rocky1 (L14 DsFi) ERROR: range check error (20 / 20) (D (Sprint)) 02:26:19 if I want to do anything saveCompat related on local that's not just following an existing saveCompat crash I got a copy of from the server. I...need to know the answer to this question: "How do I get stack trace on my platform?" 03:18:55 Unstable branch on crawl.beRotato.org updated to: 0.26-a0-393-g67388ea (34) 03:46:35 Fork (bcadrencrawl) on crawl.kelbi.org updated to: 0.22.1-2865-ge5c6407a08 04:49:58 i think it's currently not possible to tell allies to attack a new enemy ... `ta` and `tr` seem to simply not do anything. Just updated trunk to check. Sorry for being too lazy for a github issue. 04:57:37 -!- Tiobot is now known as Guest27641 06:41:12 TELL commands related to summons do nothing 13https://crawl.develz.org/mantis/view.php?id=12323 by Octopode-monk-of-XOM 08:36:24 -!- aidanh_ is now known as aidanh 08:38:47 -!- behalebabo_ is now known as behalebabo 09:02:26 aidanh: re: monitoring Crawl's disk utilization. CXC has a dashboard with a bunch of stats, including disk I/O: http://crawl.xtahua.com:19999/#menu_system_submenu_disk 09:09:10 Ge0ff: nice, that's definitely a step in the right direction 09:11:25 if you could merge that with a time-series of deployed crawl binary versions / webserver versions, it'd be very nice 09:49:46 advil: ebering: this isn't too relevant right now since post-tourney, but CDO has migrated to its new host and services should be restored 09:50:01 @?goblin 09:50:01 goblin (15g) | Spd: 10 | HD: 1 | HP: 3-5 | AC/EV: 0/12 | Dam: 4 | 10weapons, 10items, 10doors | XP: 1 | Sz: small | Int: human. 09:50:03 nice 09:50:08 wow that seemed faster than usual somehow 09:52:43 wonder if this means that trunk builds will build again 09:54:01 no 09:54:15 wasn't the os going to be updated? 09:59:40 ohh maybe I'm forgetting how this worked 09:59:57 wasn't cdo building these with some random git repo at some random commit from 10 years ago? 10:00:42 I wonder if napkin could just install the standard packages (that probably didn't exist then) and revert to a normal cross-compile 10:00:42 advil: yeah, I think there's a thing about trunk version changes that needs updating that we don't remember 10:00:42 hrm, are you talking about mingw specifically 10:00:49 I guess those are the only trunk builds cdo does aren't they 10:00:55 yes 10:00:57 and he mentioned they were broken and pasted an error 10:01:04 they have been broken for a while because the compiler is old 10:01:34 oh, ok; he showed me a compilation error on the new host, but you're saying the problem is we actually have a bad toolchain? 10:01:39 or maybe you could take a look now 10:01:42 I know what broke it, and I tried a quick fix that didn't work, but I'm tired of reverting things that are c++11 10:01:51 right, we have now... 10:02:06 yes, it's a toolchain issue 10:02:36 it's one of those slightly pre-c++11 compilers that's missing a few things, and at some point I finally added something (constexpr constructor) that broke it 10:02:40 we have 4.9.2 mingw looks like 10:03:56 advil: Napkin is on the roguelikes discord if you wanted to PM him with details; he responded to me just a few minutes ago 10:04:15 I'm still trying to figure out what this is actually using to build 10:06:20 are the standard cross-compiling tools installed somewhere on cdo? 10:08:11 ok, right, this is using a version of mingw from the following repo on a commit from 2014: https://github.com/mxe/mxe 10:12:32 it does in fact look like recent mingw cross-compiling tools are installed 10:12:48 so I can probably do this myself if I can understand the custom config on cdo 10:17:16 hm no I'm not sure why I thought that 10:22:48 advil: ah yeah, now I remember you discussing this 10:22:51 some time ago, I mean 10:23:08 I assume this comes from a time when you couldn't install something from apt that would work out of the box 10:23:26 we really should shift to doing this on ci 10:23:43 yep, would be great 10:23:54 but I think we never figured out the hosting part of it 10:24:15 if we could copy from the github CI over to CDO maybe 10:24:37 that's probably possible, could also put in object storage somewhere 10:25:09 b2 now supports s3 api so I think that would make it easier to use my b2 archive? 10:25:23 but even rotating tmp storage on github might be fine 10:25:53 I have sometimes used those builds for bisecting so that is a nice use case to keeping them around for a longer arc, but that ... might be a bit niche 10:27:13 New branch created: fix-directn-flicker (1 commit) 13https://github.com/crawl/crawl/tree/fix-directn-flicker 10:27:13 03Aidan Holm02 07[fix-directn-flicker] * 0.26-a0-390-gb3a8b14: Fix excessive redraw when autofiring ranged weapons 10(10 minutes ago, 1 file, 23+ 5-) 13https://github.com/crawl/crawl/commit/b3a8b14745a8 10:30:46 hm, this would be a good one to look at maybe but the player seems to have a custom dump order that doesn't print our data :-( https://underhound.eu/crawl/morgue/Sar/morgue-Sar-20200722-053018.txt 10:32:05 though there's about 40k turns of zigging at the end but it was still 180k turns when they started that 10:33:11 wow, Sar 10:33:18 I have sometimes actually wondered if autofight implemented in lua is more trouble than it's worth 10:34:04 it's very cleverly implemented, no shade intended 10:34:35 but this is far from the first time it's been at the very edge of what works well as a lua script 10:34:43 does seem a bit weird to have such a core feature implemented that way 10:35:14 i think it's good to implement core features in lua, otherwise you inevitably end up with an api that is broken and useless 10:35:27 right, but 10:35:36 level generation seems like an example of a good core lua feature 10:35:43 really interactive stuff though? 10:35:50 I think if I were implementing this from scratch in lua as a core feature I would not do it by sending keys to the targeter ui :-) 10:36:23 clua seems more like a nicety we provide to players that mostly about bots (excluding TAB) 10:36:29 it's not even sending keys specifically to the targeter UI, it's just pushing keys (well, commands) to the input buffer 10:36:37 vs dlua which is really about enabling creation of game content without recompilation 10:36:48 yeah, and it's a very brittle (but clever) way to do it 10:37:03 i agree with that, definitely 10:37:08 speaking from the amount of time I've spent fixing really hard-to-pin-down bugs with that aspect of it 10:37:31 i think providing an api to enqueue commands is a bad idea 10:37:33 I think in a way it's exposing that the lua targeting api is bad 10:37:49 ... what targeting api :) 10:38:42 i.e. if I wanted to implement this feature in lua now, I'd implement a real lua targeting api first 10:38:42 (I still think this one would be better as hypothetical c++ though) 10:39:16 i don't know if you're using 'targeting' in a more general way, but part of the problem is that there's no *firing* api 10:39:47 so resorting to brittle UI automation is the only possible course of action 10:53:44 aidanh: sounds like you want the clua api to be more "first class" when it comes to UI manipulation? Is there any overarching reason for that, given that we don't actually have a ton of non-tab clua UI things? 10:55:16 !lg lightli ddck -log 10:55:16 1. Lightli, XL27 DDCK, T:159313: https://crawl.kelbi.org/crawl/morgue/Lightli/morgue-Lightli-20200722-025514.txt 10:55:17 I guess if you just have good support for inspecting game data but not very good support for manipulating the UI, there's a limit to what you can really do with clua, but we don't have a ton of consumers of the API, it feels like 10:55:18 this game is sort of interesting for our data project 10:55:45 I'm not sure he's really defending this particular case 10:55:48 oh, xom scumming 10:56:06 yeah, but it actually looks to me like it would be doable within the constraints of the clock, something I was wondering about 10:56:09 yeah, that's what I was trying to get a sense of, why is the general case important 10:56:24 well, maybe 10:56:33 hrm, 85k on d:1 10:56:34 85k turns on D:1 is hard to mimic 10:57:11 I was sort of wondering what xomscumming is like if you just spread out your scumming time across floors as much as possible 10:57:35 we need a xomscumming bot 10:57:43 haha 10:57:46 I guess we have lightli... 10:58:26 gammafunk: i actually don't want ui manipulation to be allowed at all, but instead apis that expose actions on objects 10:58:38 e.g. firing a particular item at a certain location 10:58:45 it's very 2020s AI, why automate something when you can just outsource it to underpaid workers 10:59:49 heh 11:00:48 aidanh: oh, I think I get what you mean, and yeah I guess I could see that being a much less error-prone way of working with clua 11:02:16 we should just make crawl multiplayer so we'd have more of a need for bots 11:16:00 The build passed. (fix-directn-flicker - b3a8b14 #14040 : Aidan Holm): https://travis-ci.org/crawl/crawl/builds/710765696 11:17:53 ikjuyhgtrf (L13 HOFi) ERROR: range check error (20 / 20) (D (Sprint)) 11:45:04 aidanh: we *do* have something closer to a firing api 11:45:04 the targeter calls the lua hook ch_target_monster(x, y) to ask lua if this is the default target it would like 11:50:36 til 11:51:37 03Aidan Holm02 07* 0.26-a0-394-g0b5ff55: Fix excessive redraw when autofiring ranged weapons 10(2 hours ago, 1 file, 23+ 5-) 13https://github.com/crawl/crawl/commit/0b5ff556b1e2 11:52:02 see also ch_target_monster_expl and ch_target_shadow_step 11:52:50 wait, so that callback is called for each square in view with a monster in it? 11:53:21 yep 11:53:29 spiralling outwarad from the player 11:53:33 how is the callback supposed to know that it isn't just about to be invoked with an even better pick? 11:55:40 surely calling it once with a list of all possible options would make more sense 11:56:11 I think the use case envisoned is "evaluate the situation using the view.* and monster.* methods then set ch_target_monster to a closure that returns true if and only if called on the desired coordinates" 11:56:46 which, really means the callback should probably just be called to ask for its desired pick 11:56:54 also its called for all cells in sight, not just those with monsters 11:58:36 you would then have to set the callback anew each time, no? 11:58:58 or at least evaluate the situation from the callback multiple times 11:59:45 yes 11:59:59 which is why I said it should really be called once 12:00:49 yup 12:00:52 without a major overhaul to move from the enque commands model to providing an api to specific actions having a callback that you set anew to describe how an action should be taken seems to be the only reasonable route 12:01:32 hm, depends what you want to accomplish in lua 12:01:40 provinding client lua to say express "cast freezing cloud at 4,-2" 12:02:09 Unstable branch on crawl.kelbi.org updated to: 0.26-a0-394-g0b5ff556b1 (34) 12:02:21 Enabling bots/automation is one thing, but IMO the most important part of having an extensible automation library is accessibility 12:03:49 like this user's experience: https://www.reddit.com/r/dcss/comments/9c4a7j/yavp_a_voicecontrolled_mibe_of_trog_thanks_devs/ 12:04:17 I'm not sure how much lua they used vs. voiceattack 12:04:24 that's very cool 12:04:59 i don't know how you would express "cast freezing cloud at 4,-2" in a way that made use of callbacks 12:05:33 i guess setting a targeting callback to choose 4,-2, and then calling the zap command? 12:05:42 ugly 12:05:49 yep 12:06:24 also setting chk_spell_choice = spells.letter("Freezing Cloud") 12:06:37 you're kidding 12:06:43 no 12:07:33 wowzers 12:07:46 the control flow from CMD_CAST_SPELL through choosing a spell and displaying that prompt and then to the direction chooser zigzags through the codebase 12:08:39 having a theoretical lua spells.cast("Freezing Cloud", 4, -2, true) (last parameter for force/no user confirm) would just move that nightmare from a lua user's problem to the C++ codebase's problem 12:10:58 in a related but probably more tractable direction, did you get a chance to look at my cursor-placement discussion in the screenreading issue 12:11:39 i think it's better to move it to the c++, myself, no need for the guts to spill out into lua 12:12:44 i think the problem is that spells.cast(...) would be called presumably from a lua-bound function, so it's run while processing commands 12:12:55 so there's this ugly reentrancy 12:16:34 hm, interesting; implementing text-to-speech from lua might be a good fit 12:17:07 as for moving the cursor, there's some limited support for that, used to position the cursor for textbox widgets 12:17:27 Yeah, though from what I've understood from some blind players I've chatted with moving the cursor to the relevant thing to read is already an improvement 12:18:34 i wonder if it works if the cursor is invisible 12:19:39 hm, for message window prompts, i think the cursor is also positioned where one would start typing 12:20:21 !tell gammafunk my last thought on the issue, and again this is neutral because i like the spell, but frozen ramparts really seems to function like a lvl 3 freezing cloud spell spell in tight spaces. the spaces next to the wall act exactly like damaging clouds with one exception: the player can walk through it freely. just my observation 12:20:22 twelwe: OK, I'll let gammafunk know. 12:25:43 aidanh: at least with VoiceOver on mac, even with the cursor invisible when the cursor position updates it moves the VoiceOver cursor to that line and triggers it to start reading 12:26:09 this is usually the @ but not always? VoiceOver + iTerm2 seems to get confused with the two ncurses buffers and tries to read the hidden one sometimes 12:26:19 I dunno if that's our fault or the screen reader's fault 12:41:45 the two ncurses buffers? 12:42:59 for anything in sub-menus, it should be easy enough to auto-move the cursor after rendering 12:43:33 for the main game view, i'm not sure; that's the only view that's not yet widgified, due to its entanglement with the input loop in main.cc 12:44:58 where would I go looking for cursor moving in sub-menus 12:47:56 yall code optimizing today? 12:50:44 aidanh: sorry, not the two ncurses buffers, the normal and alternate screen features of my terminal emulator 12:51:13 which leads me to belive that it's iTerm2's problem implementing the accessibility api, not crawl's 12:57:48 prollumsloof (L12 HOMo) ASSERT(mons->get_foe()) in 'mon-cast.cc' at line 6378 failed. (Lair:2) 12:58:20 !crashlog 12:58:20 22401. svalbard, XL9 OpDe, T:7451 (milestone): http://crawl.develz.org/morgues/trunk/svalbard/crash-svalbard-20200721-182716.txt 12:58:20 !crashlog prollumsloof 12:58:20 No milestones for prollumsloof (crash). 12:58:46 !crashlog prollumsloof 12:58:47 No milestones for prollumsloof (crash). 12:58:49 hmph 13:12:51 Unstable branch on crawl.akrasiac.org updated to: 0.26-a0-394-g0b5ff55 (34) 13:12:57 no log for you 13:23:41 hmph 14:27:53 is that a problem on my end 14:31:45 Wereblood howls ignore form vocal nuances 13https://crawl.develz.org/mantis/view.php?id=12325 by hayenne 14:31:45 Autoexplore won't explore when unopened runed doors are on the level 13https://crawl.develz.org/mantis/view.php?id=12324 by hayenne 14:33:56 !lm prollumsloof 14:33:56 413. [2020-07-22 16:57:38] prollumsloof the Severer (L12 HOMo of Beogh) ASSERT(mons->get foe()) in 'mon-cast.cc' at line 6378 failed on turn 13789. (Lair:2) 14:34:01 !lm prollumsloof -log 14:34:01 413. prollumsloof, XL12 HOMo, T:13789 (milestone): https://crawl.kelbi.org/crawl/morgue/prollumsloof/crash-prollumsloof-20200722-165738.txt 14:34:12 !crashlog prollumsloof 14:34:13 1. prollumsloof, XL12 HOMo, T:13789 (milestone): https://crawl.kelbi.org/crawl/morgue/prollumsloof/crash-prollumsloof-20200722-165738.txt 14:34:23 maybe there was just a time lag for some reason 16:21:03 prollumsloof (L12 HOMo) ASSERT(mons->get_foe()) in 'mon-cast.cc' at line 6378 failed. (Lair:2) 16:50:30 prollumsloof (L13 HOMo) ASSERT(mons->get_foe()) in 'mon-cast.cc' at line 6378 failed. (Lair:6) 18:23:53 Unstable branch on underhound.eu updated to: 0.26-a0-394-g0b5ff556b1 (34) 18:38:14 -!- TAS-2012v is now known as TAS_2012v 18:46:36 -!- jfcaron_ is now known as jfcaron 18:46:36 -!- aidanh_ is now known as aidanh 20:03:29 Find an invisible summoner by x-ing the summon 13https://crawl.develz.org/mantis/view.php?id=12326 by Midn8 20:30:43 Maud (L20 FeAE) ASSERT(!invalid_monster(&mons)) in 'mon-death.cc' at line 1712 failed. (Vaults:1) 20:30:59 03kate-02 07* 0.26-a0-395-gfeb5eb7: Rebrace 10(4 days ago, 2 files, 20+ 9-) 13https://github.com/crawl/crawl/commit/feb5eb7db656 20:30:59 03kate-02 07* 0.26-a0-396-gfc2ff0c: Don't give Artificers branded starting weapons 10(6 hours ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/fc2ff0c6b637 20:30:59 03kate-02 07* 0.26-a0-397-gd09758e: Use form-specific shout verbs during wereblood (#12325) 10(6 hours ago, 2 files, 8+ 1-) 13https://github.com/crawl/crawl/commit/d09758e20de0 20:30:59 03kate-02 07* 0.26-a0-398-gee3a7d6: Adjust wording of Chei passives (#12321) 10(5 hours ago, 1 file, 3+ 3-) 13https://github.com/crawl/crawl/commit/ee3a7d6640a0 20:30:59 03kate-02 07* 0.26-a0-399-g6c26fa7: Clarify sleeping butterfly movement 10(46 minutes ago, 2 files, 18+ 13-) 13https://github.com/crawl/crawl/commit/6c26fa703df9 20:30:59 03kate-02 07* 0.26-a0-400-g239875a: Don't leak information about invisible summoners (#12326) 10(12 seconds ago, 1 file, 3+ 1-) 13https://github.com/crawl/crawl/commit/239875ab8653 20:32:58 kate-: you missed reupdating the artificer description 20:33:06 !crashlog Maud 20:33:06 1. Maud, XL20 FeAE, T:69738 (milestone): https://crawl.kelbi.org/crawl/morgue/Maud/crash-Maud-20200723-003032.txt 20:36:31 03kate-02 07* 0.26-a0-401-g3f6f889: Update Artificer description (ebering) 10(2 minutes ago, 1 file, 2+ 2-) 13https://github.com/crawl/crawl/commit/3f6f88936d07 20:36:31 03kate-02 07* 0.26-a0-402-g649b3d1: Further clarify Chei passive 10(68 seconds ago, 1 file, 3+ 3-) 13https://github.com/crawl/crawl/commit/649b3d1337eb 20:47:58 Unstable branch on crawl.kelbi.org updated to: 0.26-a0-402-g649b3d1337 (34) 22:20:39 Idea: artifacts with negative enchantment level should be enchantable with scrolls up to +0. 22:21:00 It would make less artifacts just immediately garbage cuz of -. 22:47:00 prollumsloof (L13 HOMo) ASSERT(mons->get_foe()) in 'mon-cast.cc' at line 6378 failed. (Lair:6) 23:18:26 prollumsloof (L13 HOMo) ASSERT(mons->get_foe()) in 'mon-cast.cc' at line 6378 failed. (Lair:6) 23:19:38 prollumsloof (L13 HOMo) ASSERT(mons->get_foe()) in 'mon-cast.cc' at line 6378 failed. (Lair:6) 23:23:43 hm that appears to be monster cloud wand usage as an ally 23:50:37 Why isn't expose to element part of ouch/hurt? All the times it's called it's called immediately after; the refactoring makes sense; though you'd have to pass the beam_type to ouch/hurt. 23:50:37 Bcadren: You have 1 message. Use !messages to read it. 23:51:42 !tell floraline In this case; my first attempt at fixing the save (which went live at rebuild this morning) fixed the save so I don't need it. But for future reference it's bhylton7@gmail.com . 23:51:42 Bcadren: OK, I'll let floraline know.