00:31:40 Fork (bcrawl) on crawl.kelbi.org updated to: 0.23-a0-4778-g1c48639af6 03:22:13 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-4778-g1c48639af6 07:18:05 <10P​leasingFungus> @advil saw an interesting bug (?) report: using ?ea/?ew with keyboard, mouse cursor selected an undesired item for enchantment 07:18:21 <10P​leasingFungus> haven’t tried to repro yet, sorry 08:04:03 -!- Stella is now known as transbee 08:51:50 is it intended that failing to evoke a scarf of invis still costs max HP? 08:52:03 it's very weird because it doesn't cost MP 08:52:29 and in general failing to activate abilities doesn't cost MP or piety etc 08:54:21 I guess it's intentional given the description says "This ability will temporarily drain your maximum hit points when used, even unsuccessfully." 08:54:26 but it seems wildly inconsistent 08:54:34 <10P​leasingFungus> yes it’s intentional 08:54:35 at least make it cost MP on failure too 08:54:40 <10P​leasingFungus> sure 08:54:44 <10P​leasingFungus> remind me 08:54:45 it's just weird to only cost one part of the cost 08:55:56 <10P​leasingFungus> alternatively i could remove the mp cost 08:56:13 that would be fine too 08:56:51 <10P​leasingFungus> 🙂 08:57:43 probably would make more sense, I think a lot of MP costs for activated abilities that aren't spammable are just around for legacy reasons at this point 08:59:05 (specifically thinking of evocable blink, assuming it has the same sort of -blink cooldown as the spell at this point) 09:00:57 <10P​leasingFungus> i don’t remember if it does 09:01:03 <10P​leasingFungus> but it could! 09:01:40 <10P​leasingFungus> vaguely think there’s something interesting about having ‘shields down’ moments and interaction with eg ghost moths 09:01:47 <10P​leasingFungus> but don’t feel strongly 09:02:37 The drain on failure was really good when it was skill drain :-( 09:03:28 I think it probably causes more confusion with the whole trog/antimagic etc interactions than it gives 09:03:58 and you could turn the ghost moth thing on its head - if you have blink spell, blink evocable is still useful if it doesn't cost MP 09:04:42 <10P​leasingFungus> sure 09:04:58 <10P​leasingFungus> and if you have invisibility spell..! 09:09:20 (I just found a +Blink ring and it does have the -blink cooldown, and also costs 1 MP) 09:11:06 <10P​leasingFungus> nice 09:16:02 -!- transbee is now known as Stella 09:29:02 <06a​dvil> yeah, I saw a screenshot, haven't looked into it 09:29:21 <06a​dvil> stupid question, is enchanting floor items supposed to be possible at all or did that get accidentally enabled? 09:29:46 <10P​leasingFungus> it seems reasonable to allow or ban 09:29:58 <10P​leasingFungus> sort of an edge use case 09:30:11 <10P​leasingFungus> vaguely think it’s intentional 09:34:30 it seems consistent with being able to identify floor items 09:37:09 -!- Yermak_ is now known as Yermak 09:40:17 <10P​leasingFungus> sure 10:08:04 <09g​ammafunk> elliptic: Yeah the failure cost is to disincentivise out of los usage with high fail rate. Likewise seems fine to me to remove MP cost, since drain is a very effective cost 11:32:05 yeah, my first reaction when I got the drain despite failing was "that's probably good", but then my second reaction was "but I still have all my MP, seems buggy" 11:33:58 btw, can we remove sinv from some of the more random things that have it now that invis is hard/painful to get? 11:34:18 like, I never remember exactly which p's in vaults have it 11:35:29 <10P​leasingFungus> i purged a bunch of them uh 11:35:33 <10P​leasingFungus> a few months ago 11:35:49 frosthearts still have sinv! 11:35:55 <10P​leasingFungus> had a commit message explaining the reasoning for remaining monsters which had it 11:36:06 <10P​leasingFungus> could certainly purge more 11:36:30 <10P​leasingFungus> frosthearts fall under the ‘wizards (most)’ category of sinv monsters 11:36:49 <10P​leasingFungus> wizards (except the ones that don’t) 11:37:23 <10P​leasingFungus> my lore is that they’re casting the see invisible spell while you’re offscreen 11:37:40 <10P​leasingFungus> keeping their enchantments and divinations running 11:38:07 <10P​leasingFungus> i am not actually wedded to this but it does make me smile 11:39:45 <10P​leasingFungus> !gitgrep 1 sinv 11:39:49 <04C​erebot> PleasingFungus * 0.28-a0-1413-g493c1fd462: Strip sInv from various enemies (8 months ago, 1 file, 20+ 20-) https://github.com/crawl/crawl/commit/493c1fd46232 11:39:56 @??wizard 11:39:56 wizard (06p) | Spd: 10 | HD: 10 | HP: 31-45 | AC/EV: 2/13 | Dam: 6 | 10weapons, 10items, 10doors, evil, spellcaster | Res: 06will(60) | Corpse | XP: 656 | Sp: fireball (3d19), banishment, blink, invisibility | Sz: Medium | Int: human. 11:40:02 <10P​leasingFungus> wow, did not expect that to work 11:40:28 <08n​icolae> lol at mlio minderbender 11:40:49 <10P​leasingFungus> Things that still have sInv: - spriggans, nagas, trolls and jellies - vault guards - "wizards" (some) - high-tier demons & all angels - some statues and some other stuff Plausible that more pruning is called for in future. 11:41:06 <10P​leasingFungus> i think i cut it from angels after? 11:41:10 frostheart is clearly a battlemage anyway 11:41:10 <10P​leasingFungus> i forget 11:41:13 and not a wizard 11:41:27 <10P​leasingFungus> frostheart counts as a vault guard? 11:41:33 <10P​leasingFungus> BATTLEMAGE 11:41:53 <10P​leasingFungus> eh, don’t mind turning that one off 11:42:04 <10P​leasingFungus> frostheart got plenty going on anyway 11:42:11 vault guards only have see invisible because they are guards IMO 11:42:14 they learn it in guard school 11:42:22 <10P​leasingFungus> agreed 11:42:49 <10P​leasingFungus> i made guardian mummies sinv and then kate (?) made me move it onto priests 11:42:50 <10P​leasingFungus> iirc 11:43:15 <10P​leasingFungus> vaguely wonder if newish non beam based spells need some extra invis fuzzing logic 11:43:36 <10P​leasingFungus> so eg frostheart can cast creepy frost when you’re not adj to a wall 11:43:36 I do think deep trolls having sinv is weird btw 11:43:40 since regular trolls don't 11:43:50 <10P​leasingFungus> @??troll 11:43:51 <04C​erebot> troll (T) | Spd: 10 | HD: 7 | HP: 32-44 | AC/EV: 3/10 | Dam: 20, 15(claw), 15(claw) | doors, regen | Res: will(20) | Corpse | XP: 304 | Sz: Large | Int: human. 11:43:54 <10P​leasingFungus> huh 11:44:00 player trolls also don't 11:44:03 <10P​leasingFungus> they used to have a sense of smell 11:44:07 <10P​leasingFungus> i think the idea was 11:44:14 <10P​leasingFungus> things living in caves 11:44:19 <10P​leasingFungus> bad eyesight good smell 11:45:52 <10P​leasingFungus> think it’s good for a few monsters each branch to have sinv 11:46:13 it's possible that deep trolls are needed for enough depths monsters to have sinv, yeah 11:46:14 <10P​leasingFungus> think it’s good for sinv to come on consistent intuitive categories 11:46:36 mainly vaults has a billion different sinv monsters and there isn't much pattern 11:46:50 <10P​leasingFungus> sure 11:46:59 <10P​leasingFungus> vaults is a silly place 11:47:07 <10P​leasingFungus> didn’t really succeed at that rework 11:47:08 depths does have spriggans and some liches and such I guess 11:47:14 <10P​leasingFungus> yeah 11:47:32 <10P​leasingFungus> spriggans might be enough 11:47:39 03robertxgray02 07https://github.com/crawl/crawl/pull/2188 * 0.30-a0-40-g83e7cc30c7: Standardize Linux packages 10(1 year, 1 month ago, 19 files, 347+ 514-) 13https://github.com/crawl/crawl/commit/83e7cc30c76b 11:47:39 03robertxgray02 07https://github.com/crawl/crawl/pull/2188 * 0.30-a0-41-gca1d0a7ddf: Fix some paths on AppImage builds GAME is used to set the name of the executable. Define SAVEDIR because the current path is used otherwise. 10(11 months ago, 1 file, 3+ 2-) 13https://github.com/crawl/crawl/commit/ca1d0a7ddfe2 11:47:39 03robertxgray02 07https://github.com/crawl/crawl/pull/2188 * 0.30-a0-42-g8629c9c750: Update AppImage and AppStream - AppImage: Fix missing fonts - AppImage: Prepare files for GitHub Actions - AppStream: Use reverse domain names as identifiers - AppStream: Add content rating 10(9 months ago, 7 files, 73+ 18-) 13https://github.com/crawl/crawl/commit/8629c9c75037 11:47:39 03robertxgray02 07https://github.com/crawl/crawl/pull/2188 * 0.30-a0-43-g005a4d490b: Configure GitHub Actions for AppImage builds 10(9 months ago, 2 files, 64+ 0-) 13https://github.com/crawl/crawl/commit/005a4d490bbb 11:47:39 03robertxgray02 07https://github.com/crawl/crawl/pull/2188 * 0.30-a0-44-gb82ad3ccd4: New XDG_NAME build variable 10(2 days ago, 5 files, 30+ 23-) 13https://github.com/crawl/crawl/commit/b82ad3ccd4fa 11:47:39 03robertxgray02 07https://github.com/crawl/crawl/pull/2188 * 0.30-a0-45-g01c43a71ca: Build debian packages using github actions 10(4 hours ago, 4 files, 101+ 3-) 13https://github.com/crawl/crawl/commit/01c43a71ca35 11:48:11 also I'm looking at this commit and who is mlio minderbender 11:48:23 ??mlio 11:48:23 mlioglotl[1/2]: An abyssal-themed unique than can use its {Corruption} to temporarily summon hostile abyss monsters and transform the surrounding area. It's fast (speed 14) and also casts {Cause Fear}. New in 0.28. 11:48:26 oh, mlioglotl 11:49:16 that unique is strange because the corruption usually doesn't do much IME 11:49:40 it's dangerous because it's a fast monster that has a bunch of health and hits fairly hard 11:50:02 and it spends a turn redecorating the dungeon a bit 11:50:48 <10P​leasingFungus> very flavour first design 11:50:51 <10P​leasingFungus> which isn’t terrible 11:50:53 like the majority of the time the corruption just doesn't summon anything worth paying attention to 11:51:08 <10P​leasingFungus> !killratio miloglotl 11:51:11 <04C​erebot> No battles for miloglotl. 11:51:12 <10P​leasingFungus> oop 11:51:21 <10P​leasingFungus> !killratio mlioglotl 11:51:25 <04C​erebot> mlioglotl wins 1.424% of battles. 11:51:31 <10P​leasingFungus> iirc took ages to get its first kill 11:51:37 <10P​leasingFungus> don’t remember if we buffed it 11:51:46 you nerfed it by removing sinv! 11:52:20 anyway I assume the corruption could be tweaked in various ways 11:52:57 one thing I was wondering about is making the corruption-summoned monsters durably summoned, so you still have to deal with them after killing mlio 11:53:33 but also just tweaking it to summon twice as many monsters or something 11:54:42 <10P​leasingFungus> durably summoned is on theme for the ability 11:54:50 yeah 11:54:54 <10P​leasingFungus> unsure whether it's too mean, but we could certainly try it 11:55:05 I assume mlio can only corrupt once per level? 11:55:10 <10P​leasingFungus> uhh 11:55:51 I don't think I would want durable summons on something that could keep on doing it, but as a one-time thing it feels okay to me 11:56:14 (players can only corrupt once per level unless that changed sometime) 11:56:18 <10P​leasingFungus> it hasn't changed 11:56:23 <10P​leasingFungus> it does make for some exciting time pressure if it can keep corrupting 12:15:00 <06a​dvil> huh, do multiple corruptions on a level work at a technical level? 12:15:07 <06a​dvil> my default guess is no 12:15:32 <06a​dvil> although, why not, maybe it'd be fine 12:18:24 <10P​leasingFungus> i don’t know why they wouldn’t 12:18:39 <10P​leasingFungus> oh, temp terrain from mlio, you mean? 12:18:44 <10P​leasingFungus> that is a bit spooky 12:20:29 <10P​leasingFungus> one last thought re sinv: some argument that eyeless stuff should have it 12:20:50 <10P​leasingFungus> that notably affects some abyss monsters, elementals, dancing weapons, etc 12:20:59 <10P​leasingFungus> but think you can flavour it however 12:25:18 <05k​ate> i always assumed not having sinv was pretty intentional choice for things like dancing weapons and elementals 12:25:55 <05k​ate> nice for there to be a reasonable way for stabby characters to kill them, since they have infinite will 12:26:54 <05k​ate> elementals less of an issue probably since they're rare anyway, but it's a pretty significant thing for dancing weapons at least 12:27:45 <06a​dvil> yeah, you definitely wouldn't it to happen again until all temp terrain is gone I think? 12:27:52 <06a​dvil> but maybe as long as that's clear it's ok 12:30:13 yeah, limiting the number of magic-immune sinv things is good 12:30:33 bonus: you get to imagine googly eyes on dancing weapons 12:30:56 (maybe you don't get to do this in tiles, I don't know how detailed the tile is) 12:31:40 <10P​leasingFungus> it’s just the weapon tile with a music note on it 12:31:51 <10P​leasingFungus> should have a googley eye on it instead 12:32:14 <10P​leasingFungus> anyway, agree with kate 12:32:20 disappointed it isn't just the mirror image of the weapon tile like in console 12:32:25 <10P​leasingFungus> and elliptic 12:32:26 <10P​leasingFungus> hah 12:33:48 <09g​ammafunk> They have have blade eyes, like how gooe have eye teeth 12:35:02 Mlio's corruption continually spreading for as long as the player is in Milo's LOS might be neat 12:35:18 assume walking tomes also have googly eyes 12:35:45 now I'm imagining a cross between a spellbook and Clippy 12:37:21 oh, tentmons is another example of something that could probably stand to lose sinv 12:40:29 <09g​ammafunk> Wow tentmons 12:40:48 <09g​ammafunk> New dcss slang 12:41:49 <09g​ammafunk> I suppose that’s true ellliptic but that’s a brutal XXX nerf 12:55:46 New branch created: pull/2707 (2 commits) 13https://github.com/crawl/crawl/pull/2707 12:55:46 03Michael Del Monte02 07https://github.com/crawl/crawl/pull/2707 * 0.30-a0-4-g2eeeca803e: Add `view.timed_portal` lua function. 10(6 days ago, 1 file, 22+ 0-) 13https://github.com/crawl/crawl/commit/2eeeca803e50 12:55:46 03Michael Del Monte02 07https://github.com/crawl/crawl/pull/2707 * 0.30-a0-5-g237bddf2a6: remove non-timed portals 10(9 minutes ago, 1 file, 2+ 1-) 13https://github.com/crawl/crawl/commit/237bddf2a6fa 13:23:45 isn't that really old slang 13:24:23 and yeah, I didn't think about the whole sinv summons situation (which IMO isn't ideal - I know it's quite confusing to players) 13:27:14 <06a​dvil> man, cao import flies on arm/py3/ssd 13:27:32 <06a​dvil> *cao scoring improt that is 13:28:06 <06a​dvil> it's estimating about 45 minutes for the entire db load (minus non-available logfiles that I haven't bothered to download), used to be days/weeks 14:14:56 -!- TAS-2012v is now known as TAS_2012v 14:52:32 SlimeGoblin (L2 TrBe) Crash caused by signal #6: Aborted (D:2) 15:02:54 <09g​ammafunk> I assume you have the Sequell mirror url that should have the archives for offline servers? 15:07:34 <06a​dvil> yeah, I have them all somewhere 15:07:44 <09g​ammafunk> And yeah can we just move cao to Advil’s desktop :gammafHeh: 15:08:26 <09g​ammafunk> Official server on OS X 15:08:48 <09g​ammafunk> If it’s still called that 15:10:09 <09g​ammafunk> macOS apparently 15:55:26 <06a​dvil> ok about an hour in the end, Done processing 7.319GB from 13367110 lines; 3212 lines/s in 1:09:21. 15:55:30 <06a​dvil> !lg * 15:55:33 <04C​erebot> 15140324. beargit the Chopper (L5 MeBe of Trog), slain by a gnoll (a +3 halberd) on D:3 on 2022-09-04 22:54:10, with 93 points after 955 turns and 0:04:14. 15:55:50 <06a​dvil> stil short about 2M from somewhere 15:56:23 <06a​dvil> I noticed that LLD is misconfigured in CAO scoring, I guess its log loc moved at some point 16:11:01 <06a​dvil> it remains unfortunate that cao scoring streaks are so brittle to calculate 16:11:17 <06a​dvil> I wonder if it would be better for it not to show cross-server (or even cross-version) streaks? 16:11:57 <06a​dvil> I realize some famous streaks currently on the leaderboard would get dropped by the first, and a lot might get dropped by the second 16:16:37 in what way exactly are they brittle to calculate? 16:17:27 <06a​dvil> cao scoring stores only wins, not losses (aside from recent ones), so it really relies on reading log lines in temporal order 16:17:40 <06a​dvil> if it gets all the data at once, and correctly, that's fine 16:19:04 <06a​dvil> and the odds of old data messing it up are low, but e.g. when I fix this LLD issue it's going to get a whole bunch of old log lines so if there's any cross-server streaks that would be impacted by LLD games, they could in principle end up screwy 16:19:15 <06a​dvil> (which I think is pretty unlikely in this case?) 16:19:25 I seem to recall past discussion about fragility with respect to overlapping games cross-server 16:19:30 <06a​dvil> but it's just annoying from a maintenance perspective 16:22:23 <06a​dvil> sometimes wonder why this couldn't just be done directly on the sequell db 16:24:23 <06a​dvil> (well, aside from the obvious fact that no one has access to that and it would require a scoring page rewrite) 16:40:09 doesn't CAO scoring already store streakbreaker games? 16:40:40 in principle if it does that for every streak then it would be possible to reconstruct things 16:41:01 certainly still annoying to deal with processing things out of temporal order though 16:44:46 <06a​dvil> so the case that it really can't reconstruct is if it reads a streakbreaker game but doesn't have the streak yet, enough time/games pass that it no longer has it in the db, and then reads the sequence of wins that that streakbreaker is relative to. (I'm also pretty sure that it doesn't have any code to handle the case where it has an existing streak in the db and reads a streakbreaker later, which is perhaps more common -- but this could 16:44:46 be handled) 16:45:19 I guess that's true, yeah 16:45:24 <06a​dvil> I don't see a way to solve that without storing every game (which in 2022 is not so completely out, but I'm not sure CAO in particular is up to it) 16:46:25 in principle you don't need to store very much about the games to just compute streaks 16:46:31 <06a​dvil> yeah, I was thinking that 16:46:45 just win/loss, start/end times, and I guess some identifying info that can be used to produce a morgue link 16:46:48 <06a​dvil> just the gid data basically 16:47:16 !lg * 16:47:18 15140539. Isauros the Skirmisher (L9 MiHu of Okawaru), demolished by a komodo dragon (kmap: beast_lair_carwin_7) on D:8 on 2022-09-04 23:46:32, with 3173 points after 9618 turns and 0:51:47. 16:47:28 <06a​dvil> this is all a bit theoretical but these cases have come up from time to time (e.g. MRG's streak iirc has breaking games in the CJR data set, I needed to rerun the whole import to fix it) 16:49:59 <06a​dvil> right now I think cao is storing about 1M games in the recent games table, and all the wins ofc 16:51:59 <06a​dvil> though, probably other stuff I should worry about if I'm going to work on scoring again 16:54:33 <06a​dvil> like the fact that best players and all players are a 30M+ files 17:24:30 <09g​ammafunk> Unrelated but when I started tournament-enabled qw I ran it at its usual 10ms delay I use on cdi 17:26:33 <09g​ammafunk> Then I saw that qw had played something like 1/3 of all tournament games thus far (at like day 3) by dying lots of times with TeWr and tournament player page updates were taking 3s per page 17:27:09 <09g​ammafunk> So that spooked me real good and I set delay to 150ms instead 17:28:26 <09g​ammafunk> Going forward to help avoid game db inflation I’ll keep it at its current 100ms speed (on the bot install at least) 17:28:58 <06a​dvil> lol, that's a lot of games 17:29:15 <09g​ammafunk> !gamesby qw tewr 17:29:16 <04C​erebot> qw (tewr) has played 3638 games, between 2022-08-28 02:30:30 and 2022-08-29 02:50:06, won 0, high score 537533, total score 3826824, total turns 10062960, play-time/day 9:09:34, total time 18:19:09. 17:29:21 <09g​ammafunk> Yeah 17:29:36 <06a​dvil> well, I have updated scoring to be py3 compatible, though in characteristic fashion unfortunately the version of mysql on cao is too old for that to work 17:29:51 <09g​ammafunk> Haha 17:30:18 New branch created: hermitcraft (1 commit) 13https://github.com/crawl/crawl/tree/hermitcraft 17:30:18 03hellmonk02 07[hermitcraft] * 0.30-a0-51-gc4af336a4e: feat: New early unrand, the Hermit's pendant 10(4 minutes ago, 5 files, 52+ 1-) 13https://github.com/crawl/crawl/commit/c4af336a4eb1 17:30:24 <09g​ammafunk> I guess no word from rax about that dist upgrade? 17:30:47 <06a​dvil> no, I last pinged early or mid-summer and got nothing back 17:36:59 04Build failed for 08hermitcraft @ c4af336a 06https://github.com/crawl/crawl/actions/runs/2990112739 17:39:45 03hellmonk02 07[hermitcraft] * 0.30-a0-52-g12af3a85da: temporary tile for hermit pendant 10(61 seconds ago, 1 file, 0+ 0-) 13https://github.com/crawl/crawl/commit/12af3a85dab6 17:43:56 03hellmonk02 07[hermitcraft] * 0.30-a0-53-g81fa1cafc4: checkwhite 10(4 minutes ago, 1 file, 0+ 1-) 13https://github.com/crawl/crawl/commit/81fa1cafc4ed 17:43:56 03hellmonk02 07[hermitcraft] * 0.30-a0-54-g128760296f: add 5 mp 10(87 seconds ago, 1 file, 2+ 1-) 13https://github.com/crawl/crawl/commit/128760296fb3 18:13:11 04Build failed for 08hermitcraft @ 12af3a85 06https://github.com/crawl/crawl/actions/runs/2990139560 18:20:39 <08n​icolae> what about the kermit's pendant 18:20:43 <08n​icolae> let's you hop like a barachim 18:22:09 <09g​ammafunk> Then you’d need a barachi muppets go to Panhattan vault 18:22:26 <09g​ammafunk> But of course that goes without saying 20:44:57 <10P​leasingFungus> @Sastreii if you’re in the mood for tiling, hellmo is making a ‘hermit’s pendant’ - unrand that sets invo = 14 and evo = 0 21:44:54 New branch created: drac-me (1 commit) 13https://github.com/crawl/crawl/tree/drac-me 21:44:54 03PleasingFungus02 07[drac-me] * 0.30-a0-50-g693cfe6635: New draconian tiles (Sastreii) 10(6 minutes ago, 48 files, 28+ 88-) 13https://github.com/crawl/crawl/commit/693cfe663526 21:51:12 03PleasingFungus02 07* 0.30-a0-51-g784eafce7f: Clarify effects of Strength (acrobat) 10(70 seconds ago, 1 file, 2+ 1-) 13https://github.com/crawl/crawl/commit/784eafce7f83 21:57:29 03PleasingFungus02 07[hermitcraft] * 0.30-a0-55-ga911d89431: Tweak hermit's pendant description 10(22 seconds ago, 1 file, 4+ 4-) 13https://github.com/crawl/crawl/commit/a911d894317b 22:07:17 Unstable branch on crawl.kelbi.org updated to: 0.30-a0-51-g784eafce7f (34) 22:20:37 04Build failed for 08drac-me @ 693cfe66 06https://github.com/crawl/crawl/actions/runs/2990823616 22:27:27 -!- dulot_34_ is now known as dulo_t-34 22:34:48 Unstable branch on crawl.develz.org updated to: 0.30-a0-51-g784eafce7f (34) 22:55:50 <13S​astreii> Added to the ASAP list 🙂 22:57:08 Windows builds of master branch on crawl.develz.org updated to: 0.30-a0-51-g784eafce7f 23:14:39 Unstable branch on cbro.berotato.org updated to: 0.30-a0-51-g784eafce7f (34) 23:54:04 Monster database of master branch on crawl.develz.org updated to: 0.30-a0-51-g784eafce7f