00:47:13 03CrawlOdds02 07* 0.35-a0-1218-gca76fceb61: Stop foxfires and stars from hitting themselves 10(12 hours ago, 1 file, 10+ 11-) 13https://github.com/crawl/crawl/commit/ca76fceb61b0 00:47:13 03CrawlOdds02 07* 0.35-a0-1219-gc3ac0930b1: Don't do knockback from reflected stars 10(12 hours ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/c3ac0930b185 00:52:50 kingdomoftheworm (L2 OpHu) ERROR in 'fight.cc' at line 570: the bat trying to attack a dead your icy jade (D:1) 01:03:14 04Build failed for 08master @ c3ac0930 06https://github.com/crawl/crawl/actions/runs/37279663411 02:45:22 New branch created: pull/5486 (1 commit) 13https://github.com/crawl/crawl/pull/5486 02:45:23 03Sangdon park02 07https://github.com/crawl/crawl/pull/5486 * 0.35-a0-1220-gd50bd3a617: Add opt-in server policies for Lua RNG access and level generation 10(6 minutes ago, 3 files, 117+ 0-) 13https://github.com/crawl/crawl/commit/d50bd3a6175e 03:35:55 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-5261-gd9800d219b 03:45:35 <12g​e0ff> /me notices a bunch of new kate's commits rushes to Discord to post the Star Wars quote hellmonk already done that 😔 03:51:11 <12g​e0ff> also, re: the above PR5486: > User Lua currently has access to engine RNG helpers, which can be used to predict later outcomes. Spell-path queries can also reveal terrain the player has not seen. 03:51:13 <12g​e0ff> If this is true, this should be fixed directly instead of adding the "don't let the user cheat" compilation options 04:29:26 <11O​dds> Yeah, I’ll ask about the details of the issues being addressed here 04:59:45 Angmar (L9 DsCK) Crash caused by signal #6: Aborted (D:7) 05:12:05 <11O​dds> !crashlog angmar 05:12:11 <04C​erebot> 5. Angmar, XL9 DsCK, T:5734 (milestone): https://crawl.akrasiac.org/rawdata/Angmar/crash-Angmar-20261005-115938.txt 05:23:01 <11O​dds> Commented on the PR - these are very real issues (I just whipped up a macro to map an unseen level), which we can fix in better ways. Probably more interesting problem is spell-path queries, which we allow to use any source and destination and can thereby reveal terrain. I'm not sure what the best way to restrict this is - one thing that would work is to restrict the source/dest so that one of them is the player and the other is 05:23:01 somewhere the player can see (like for a targetter beam). But I wonder if there's a logic that locks things down less while still fixing the leak. 06:55:47 <09g​ammafunk> even those source/dest guarantees don't fully solve the information leak issues. There can be unseen areas in player los, after all. I'd think the beam resolution should be returning a "best guess" path that uses only player map knowledge, something along those lines. Maybe simply returning nil when any point of the calculated path hits an unseen tile would be enough. Would need to look a bit more into things 07:01:15 <11O​dds> I think the source dest thing nearly solves the issue (if you can see a cell right now you can see a path there), and to the extent it doesn’t I think targeters would have the same leak? 07:01:48 <11O​dds> Yeah beam resolution (anywhere) using map knowledge would be ideal but also I suspect much more work 07:21:35 <11O​dds> Hmmm. Maybe cases like bouncing still leak even if you are firing at visible targets. So yeah, we probably need to do something like stop the path when it goes out of the player's current vision. Doing only that still does have leaks - because ray choice between seen points can depend on unseen terrain - but I think that doing both that and forcing the player to be the source would work. 07:57:19 <04d​racoomega> To be clear, are you saying that unseen terrain can leak out of LoS info for beams from the player to a seen location only via lua api or also from the in-game targeter? 07:58:24 <11O​dds> Only the Lua API - and the leak is broader than that (it can leak unseen terrain for a beam anywhere, i.e. it can map the whole level) 07:59:42 <04d​racoomega> No, I gathered that part. But you saying that there would still be leaks even if restricted to aiming at places you could aim at with the in-game targeter 08:00:26 <11O​dds> Yes, because I think in some cases the beam might then go out of your vision (which the targeter wouldn't show) 08:01:06 <11O​dds> E.g. if you do a bounce shot round a corner, that might aim within your vision but then reveal tiles outside it. Or shooting through fog clouds. 08:01:55 <04d​racoomega> Is there any reason this API shouldn't just be querying a targeter? 08:02:45 <11O​dds> If we restrict it so that the player has to be the origin of the spell, I can't think of one 08:02:47 <04d​racoomega> (Which may or may not themselves sometimes have 'leaks', but at least lua isn't revealing anything you couldn't know already and the information is visible) 08:03:07 <04d​racoomega> Much easier to find and patch those leaks also, if they happen 08:03:19 <04d​racoomega> Since regular players will encounter them 08:03:55 <11O​dds> Yeah, as long as the thing the lua returns is exactly the thing shown in the game 08:05:09 <11O​dds> (I think targeter beams actually do calculate this same leaky path but then just doesn't draw the bits you shouldn't see) 08:29:21 03DracoOmega02 07* 0.35-a0-1220-g0f941a9d54: Add tracking for percentage of each XL spent in a given form 10(7 minutes ago, 7 files, 138+ 2-) 13https://github.com/crawl/crawl/commit/0f941a9d54d5 08:29:21 03DracoOmega02 07* 0.35-a0-1221-gea32904655: Make the player actually require the claimed 10 XP for XL 2, instead of 9 10(6 minutes ago, 1 file, 2+ 2-) 13https://github.com/crawl/crawl/commit/ea3290465564 08:45:47 <09h​ellmonk> Huge player nerf 09:11:58 someone will die to a quokka because of this 09:21:00 robertstho (L19 DjFE) Crash caused by signal #11: Segmentation fault (Shoals:1) 09:21:34 !crashlog robertstho 09:21:36 17. robertstho, XL10 CoMo, T:9625 (milestone): https://crawl.akrasiac.org/rawdata/robertstho/crash-robertstho-20250926-190455.txt 09:22:19 !crashlog robertstho xl=19 djfe place=shoals:1 09:22:21 1. robertstho, XL19 DjFE, T:27926 (milestone): https://cbro.berotato.org/morgue/robertstho/crash-robertstho-20261005-162050.txt 09:22:33 (tickticktick…) 09:23:57 uhh… targeter crash 10:58:49 <08o​____0> spell=SPELL_RIMEBLIGHT 11:10:19 <08o​____0> pos = {x = 0, y = 22007} ah, see that feels wrong 😅 11:19:29 <04d​racoomega> Yeah, I see what the issue is 11:20:00 <04d​racoomega> %git 16ceb2c 11:20:01 <04C​erebot> DracoOmega * 0.35-a0-1122-g16ceb2c681: Remove an awkward hack involving bolt.path_taken (4 weeks ago, 4 files, 11+ 15-) https://github.com/crawl/crawl/commit/16ceb2c68134 11:20:06 <04d​racoomega> (I guess I missed at least one) 11:20:41 <08o​____0> ah so the pos was just junk data because the array was empty 11:20:58 <04d​racoomega> (A beam with a 0-length path, probably because you aimed it at wall) 11:21:32 <08o​____0> Looks like the statue being in the way but force casting anyway causes it 11:22:12 <04d​racoomega> Looking over target.cc, this may be the only one that does something like this 11:27:43 03DracoOmega02 07* 0.35-a0-1222-gbabda3026e: Fix a crash when aiming Rimeblight at a wall (rypofalem) 10(71 seconds ago, 1 file, 3+ 0-) 13https://github.com/crawl/crawl/commit/babda3026e58 11:48:55 <08o​____0> I didn't even do anything lol 11:51:28 <04d​racoomega> I mean, you pointed me in the right direction! 11:52:00 <04d​racoomega> (I'd glanced over the crashlog earlier, but didn't notice the wildly wrong pos, which really was quite helpful) 12:28:08 <09g​ammafunk> hrm, I do need a solution that can use arbitrary start/end positions. I guess I might need to look into fixing this myself 12:28:49 <09g​ammafunk> one of the basic things I need is the path of e.g. a monster attempting to fire a penetrating bolt from posa to posb 12:29:09 <09g​ammafunk> I think monsters also use targeters in this case 12:37:07 <04d​racoomega> Typically they use actual tracers and we don't care about leaking a variety of things to them 15:03:38 New branch created: pull/5489 (1 commit) 13https://github.com/crawl/crawl/pull/5489 15:03:39 03pocketzot02 07https://github.com/crawl/crawl/pull/5489 * 0.35-a0-1223-g36d8e049e3: Prevent form XP tracking writing out of bounds 10(42 minutes ago, 1 file, 2+ 1-) 13https://github.com/crawl/crawl/commit/36d8e049e30b 15:18:27 <11O​dds> Arbitrary including not-seen-by-the-player? 15:23:48 <11O​dds> (On the other bug in that pr, my intention is just to make user lua use a separate rng that doesn’t depend on the game seed) 15:30:32 03pocketzot02 {DracoOmega} 07* 0.35-a0-1223-gb005dc6fb0: Prevent form XP tracking writing out of bounds 10(69 minutes ago, 1 file, 2+ 1-) 13https://github.com/crawl/crawl/commit/b005dc6fb050 15:43:31 Unstable branch on underhound.eu updated to: 0.35-a0-1222-gbabda3026e (34) 16:37:11 03hellmonk02 {DracoOmega} 07* 0.35-a0-1224-g9d3ea95bc5: Split up Ash curses 10(7 months ago, 2 files, 118+ 28-) 13https://github.com/crawl/crawl/commit/9d3ea95bc574 16:37:11 03DracoOmega02 07* 0.35-a0-1225-ga6c7512026: Split apart the Devices curse into Evocations and Shapeshifting 10(6 hours ago, 2 files, 13+ 2-) 13https://github.com/crawl/crawl/commit/a6c751202697 16:37:11 03DracoOmega02 07* 0.35-a0-1226-g80afe4ae75: Make Ash curse skill enhancements more concise 10(28 minutes ago, 4 files, 14+ 15-) 13https://github.com/crawl/crawl/commit/80afe4ae7597 16:37:11 03DracoOmega02 07* 0.35-a0-1227-ga7b1f61492: Tweak some (arguably) vestigial Ashenzari flavour 10(20 minutes ago, 13 files, 67+ 64-) 13https://github.com/crawl/crawl/commit/a7b1f614921e 16:37:11 03hellmonk02 {DracoOmega} 07* 0.35-a0-1228-g5b3774d50e: Pull some gnoll xp into manuals 10(8 days ago, 1 file, 9+ 0-) 13https://github.com/crawl/crawl/commit/5b3774d50e70 16:37:11 03Draconius02 {DracoOmega} 07* 0.35-a0-1229-g830294a5c2: Adjust Gnoll Weapons, Shapeshifting, and Invocations Aptitudes Upward +1 10(5 days ago, 1 file, 10+ 10-) 13https://github.com/crawl/crawl/commit/830294a5c2d8 17:02:06 04Build failed for 08master @ 830294a5 06https://github.com/crawl/crawl/actions/runs/37389445125 17:17:39 <09g​ammafunk> for my purposes, I'd need that a starting or ending point will be in the player's map memory, but might not currently be in LOS of the player 18:17:32 03dolorous02 07* 0.35-a0-1230-g309cdfebc1: Fix spacing. 10(26 minutes ago, 2 files, 2+ 2-) 13https://github.com/crawl/crawl/commit/309cdfebc117 18:36:55 <06d​olorous_84348> I'm looking at #5491. The problem appears to be that, in throw.cc, fire_target_behaviour::get_monster_desc looks only at the mon-info for a monster and not the actual monster, and mon-info doesn't seem to have a way to indicate that an actual monster is immune to poison (as Hell Sentinels specifically are, but demons in general are not), so the description says that Hell Sentinels can be blinded by atropa darts when it 18:36:55 should say that they're immune to all darts (since all darts are poisoned with something), according to the logic in ranged_attack::dart_check(). However, mon-info is rather a mess, and I'm uncertain of the proper fix. I could add a poison_immune boolean to it to cover this case, but it would overlap with the standard checks for individual monsters' having poison resistance of 3 or greater. 18:36:59 <06d​olorous_84348> !issue 5491 18:37:01 <04C​erebot> https://github.com/crawl/crawl/issues/5491 19:46:39 <04d​racoomega> Doesn't get_resist(mi.mresists, MR_RES_POISON) >= 3 work here? 20:03:10 <06d​olorous_84348> That does work, yes. I guess I missed it when searching (grepping for "resist" spits out a lot). Thank you. 20:03:24 <04d​racoomega> Happy to help 20:04:51 03dolorous02 07* 0.35-a0-1231-gb9ee7e9d76: Fix check for darts&poison immunity (DracoOmega). 10(6 minutes ago, 1 file, 5+ 1-) 13https://github.com/crawl/crawl/commit/b9ee7e9d764b 20:06:00 <04d​racoomega> (I mean, I am forever complaining about the amount of duplication of functionality and extra hoops monster_info sometimes results in, but this does happen to be something we can query straightforwardly) 20:06:11 <04d​racoomega> (Also, the Undead/Nonliving check is redundant now) 20:06:25 <04d​racoomega> Since they always have poison immunity 20:07:39 <06d​olorous_84348> I wondered. I'll fix that momentarily; thanks for the info. 20:15:41 03dolorous02 07* 0.35-a0-1232-g1db5de54e0: Remove now-redundant holiness checks (DracoOmega) 10(5 minutes ago, 1 file, 0+ 2-) 13https://github.com/crawl/crawl/commit/1db5de54e012 21:04:42 New branch created: pull/5492 (1 commit) 13https://github.com/crawl/crawl/pull/5492 21:04:43 03Draconius02 07https://github.com/crawl/crawl/pull/5492 * 0.35-a0-1230-gdb978c8609: Fix Jade Mantle skill enhancement message 10(4 minutes ago, 1 file, 2+ 0-) 13https://github.com/crawl/crawl/commit/db978c8609fd 21:07:41 <12e​bering> @hellmonk ash bans shapeshifting. simple 21:07:50 <12e​bering> "bound in your form" 21:07:51 <09h​ellmonk> eh 21:08:03 <12e​bering> i have been gone for a long time this is a oneoff idea 21:10:06 03Draconius02 {DracoOmega} 07* 0.35-a0-1233-g096219ce5d: Fix Jade Mantle skill enhancement message 10(10 minutes ago, 1 file, 2+ 0-) 13https://github.com/crawl/crawl/commit/096219ce5dbd 22:37:34 Unstable branch on crawl.develz.org updated to: 0.35-a0-1233-g096219ce5d (34) 22:44:43 03DracoOmega02 07* 0.35-a0-1234-gf68bec49c9: Fix icy dragon veins accidentally targeting most instead of least injured 10(31 seconds ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/f68bec49c929 23:03:37 Windows builds of master branch on crawl.develz.org updated to: 0.35-a0-1234-gf68bec49c9 23:35:58 Bnni (L27 PoHu) ASSERT(have_passive(passive_t::detect_portals)) in 'main.cc' at line 2552 failed. (Pan) 23:36:07 Bnni (L27 PoHu) ASSERT(have_passive(passive_t::detect_portals)) in 'main.cc' at line 2552 failed. (Pan) 23:36:20 Bnni (L27 PoHu) ASSERT(have_passive(passive_t::detect_portals)) in 'main.cc' at line 2552 failed. (Pan) 23:41:34 Unstable branch on cbro.berotato.org updated to: 0.35-a0-1234-gf68bec49c9 (34) 23:52:11 <11O​dds> Cool, I’ll have a look into how hard it would be to solve this using map knowledge in arbitrary places 23:59:38 Monster database of master branch on crawl.develz.org updated to: 0.35-a0-1234-gf68bec49c9