00:32:38 03CrawlOdds02 07* 0.35-a0-731-ge0f07281b7: Fix an unlinked item message 10(22 minutes ago, 1 file, 9+ 5-) 13https://github.com/crawl/crawl/commit/e0f07281b7b7 00:32:38 03CrawlOdds02 07* 0.35-a0-732-g394360a2de: Don't autoselect weapon when character is random 10(8 minutes ago, 3 files, 15+ 0-) 13https://github.com/crawl/crawl/commit/394360a2de8e 01:49:29 03CrawlOdds02 07* 0.35-a0-733-gd2521a539a: Fix tooltips on relayout 10(21 minutes ago, 1 file, 3+ 2-) 13https://github.com/crawl/crawl/commit/d2521a539ae3 03:34:04 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-5261-gd9800d219b 04:52:55 03CrawlOdds02 07* 0.35-a0-734-g743f274a0e: Fix a footgun with fear/mesmerism durations 10(32 minutes ago, 1 file, 12+ 4-) 13https://github.com/crawl/crawl/commit/743f274a0e9b 05:08:04 04Build failed for 08master @ 743f274a 06https://github.com/crawl/crawl/actions/runs/29740089751 08:44:19 <04dracoomega> I seem to be on another one of those 'simple refactoring unearths a nesting doll of bugs (that are actually more tricky than the original goal of the task)' things 08:44:45 <04dracoomega> Most of them admittedly niche, but I'm not going to just ignore them while I'm here 08:45:35 <04dracoomega> (The simple version is that a lot of warning prompts for attacks may be wrong if a coglin has a cleaving and non-cleaving weapon equipped at the same time, and also issues with certain unrands while you have the cleaving status) 09:03:28 <11Odds> Is this to do with actor::weapon() being a slightly odd idea when actors have multiple weapons? 09:05:14 <11Odds> (The monster version is especially exciting, defaulting to a random weapon for dual wielders to keep callers on their toes) 09:20:30 <04dracoomega> I mean, weapon() at least gives a consistent result for the player, but some of these codepaths don't really understand the idea that the player might have more than one weapon, and depend entirely on the properties of the first one 09:23:57 <04dracoomega> Also a handful of generally simpler/less important bugs, like how the introduction of coglins made pressing v with a cleaving weapon stop hilighting everything in cleave range, even if you have two axes. (But did anyone even use that functionality? Seems relatively unlikely, though the code attempting to make this happen ought to at least work, imo) 09:26:25 <11Odds> I'm meanwhile chasing down some nasty interlevel pathfinding bug to do with transporters 09:26:54 <11Odds> Looks like travel.cc is very much assuming the player is on the level it is reasoning about, and this is not always true 09:27:31 <04dracoomega> Huh, really? That feels like such a basic element of interlevel pathfinding that I assumed it must have accounted for it 09:28:57 <11Odds> Specifically, I'm nearly sure that what's happening is that during load_level there's a point where we save the old level while the player is on the new one, updating the travel cache using the wrong transporters 09:29:29 <11Odds> (So I think it's wrong when interlevel travel comes to use it, rather than interlevel travel itself being wrong) 09:31:36 <04dracoomega> Ah, that makes some sense 09:31:56 <04dracoomega> There have been occasional bugs with level save assuming the player is on a floor before 09:32:58 <11Odds> (I hit this repeatedly yesterday in a game with that transporter-only Zot entrance, trying to go to shops and such beyond it) 09:33:34 <04dracoomega> %git 5f9b90c 09:33:35 <04dracoomega> eg: 09:33:35 <04Cerebot> DracoOmega * 0.34-a0-1004-g5f9b90cd0b: Fix being able to freeze time on other floors by peeking at them (Darby) (9 months ago, 1 file, 5+ 4-) https://github.com/crawl/crawl/commit/5f9b90cd0b5c 09:34:59 <11Odds> I think in this case we can just fix up the travel code to not rely on this 11:56:44 03CrawlOdds02 07* 0.35-a0-735-g13275d6dd9: Fix navigation through off-level transporters 10(4 hours ago, 6 files, 84+ 44-) 13https://github.com/crawl/crawl/commit/13275d6dd930 12:06:01 <09gammafunk> integrating transporters into travel was one of those things that I thought I really wanted when I first designed the feature, since I did imagine a lot of uses where it might be required for navigation. But when I settled on the idea that monsters would not be able to take them spontaneously and it became more clear that they were best used in the context of special vaults, I sort of regretted adding the complexity of having them 12:06:01 integrated into travel pathfinding at all 12:06:22 <09gammafunk> not surprised there were some fundamental problems with said integration as travel calculations are inherently tricky 12:07:02 <09gammafunk> I guess there's a silver lining where we are starting to see a few more not-a-trap-vault uses of transporters and maybe now most of the significant bugs are ironed out 12:07:28 <09gammafunk> so perhaps the pain of implementing/fixing/maintaining that portion of code is or will be worth 12:07:32 <11Odds> I quite like the circle-transporter vaults 12:07:49 <11Odds> TBH I dont blame transporters for this bug 12:07:51 <09gammafunk> they're certainly cute 12:08:15 <11Odds> Many parts of the travel code assume(d) the player is on the level being travelled on, it was very reasonable for the transporters to do that too 12:09:46 <09gammafunk> yeah, I guess I didn't have to (or didn't think to) touch anything related to off-level travel when integrating transporters into e.g. travel.cc 12:10:14 <09gammafunk> it was more making the floodfill work when you could now have a location that kind of violated the chebychev grid 12:10:33 <09gammafunk> this one square now additionally has an exit point that's another arbitrary square on the map 12:10:33 <11Odds> Yeah, the bug here is morally in save/loading level which broke the contract. It's just that it was easier to update the contract than fix that 🙂 12:10:51 <09gammafunk> but as I recall augmenting that didn't case any massive issues 12:11:54 <11Odds> Yeah seems to be mostly working well (aside from that old minor autoexplore-through-transporters bug I fixed the other day) 12:12:16 <11Odds> Which was just a mismatch between floodfilling through transporters and not doing that in different steps 12:12:37 <09gammafunk> right, I saw that commit, I should read the diff 12:12:47 <09gammafunk> one thing that's only tangentionally related that I'll bring up since maybe you've not thought about this and are interested 12:13:36 <09gammafunk> the transporter UI has a kind of fundamental problem where we should always reveal the exact destination of a transporter in the UI when a transporter and its landing site pair have been seen (regardless of LOS_NO_TRANS, simply seen) 12:14:42 <09gammafunk> it would be ideal if this was tracked in the player travel cache and there were keys for x/X mode that let a player see a transporter/landing site at a given location (e.g. if the cursor is on either the transporter or the landing site) 12:15:16 <11Odds> Right yeah, I wondered if X and > might do this but it went somewhere pretty random 12:15:25 <09gammafunk> it's often "obvious" given how transporters are placed, but in some places it's very much not 12:15:39 <11Odds> Aren't there some vaults that rely on this not being true? Like Nem coinflips? 12:15:46 <09gammafunk> although as I say this I realize that some vaults might see this as a feature; the player doesn't know ahead of time 12:15:54 <09gammafunk> right, yeah that vault is a good example 12:16:07 <09gammafunk> not sure how to reconcile that, it's like a solitary case almost 12:16:15 <11Odds> (Also we'd have to make sure we didn't end up autoexploring through portals 🙂 ) 12:16:24 <09gammafunk> right that shouldn't be related to this 12:16:41 <09gammafunk> this is not info that should affect travel cache's dependence on visitation 12:17:08 <11Odds> But yeah this makes sense for every transporter that doesn't very deliberately hide this 12:17:21 <09gammafunk> (and tbh even that rule does trip people up when they enter a transporter vault, abandon it either via teleport or early exit, leaving monsters inside, and then are shocked to have travel magically take them back inside) 12:17:49 <09gammafunk> yeah I don't know if you can realistically implement said UI and allow for some transporters to have hidden destinations 12:18:08 <09gammafunk> I guess there's technically no reason why a transporter couldn't have a special DES mask in that regard 12:18:43 <09gammafunk> but it'd be a bit icky to have players conditioned to think transporters reveal their destination without having taken them...except for this one transporter in this one vault 12:18:56 <09gammafunk> I suppose it could have a custom tile/feature description 12:19:03 <09gammafunk> a bit like how we have for tomb hatches 12:19:25 <09gammafunk> it's a bit wonky to have a variant with such a subtle difference but it could work I guess 12:20:17 <09gammafunk> anyhow those are my transporter thoughts, maybe I'll use some in my gulch map to atone for my transporter sins 12:20:47 <11Odds> Yeah, makes sense 12:20:57 <11Odds> You could just have random-transporters not go anywhere on > in X 12:21:03 <09gammafunk> yep 12:21:20 <11Odds> (Until visited, then they should work) 14:48:35 03CrawlOdds02 07* 0.35-a0-736-g4e0084fca1: Fix a skill menu bug with console leaking into webtiles 10(61 minutes ago, 2 files, 27+ 24-) 13https://github.com/crawl/crawl/commit/4e0084fca19b 15:43:40 Unstable branch on underhound.eu updated to: 0.35-a0-736-g4e0084fca1 (34) 17:13:12 -!- Discord|2 is now known as Discord| 18:28:39 -!- MakMorn_ is now known as MakMorn 22:36:01 Unstable branch on crawl.develz.org updated to: 0.35-a0-736-g4e0084fca1 (34) 22:59:58 Windows builds of master branch on crawl.develz.org updated to: 0.35-a0-736-g4e0084fca1 23:34:55 Unstable branch on cbro.berotato.org updated to: 0.35-a0-736-g4e0084fca1 (34) 23:56:16 Monster database of master branch on crawl.develz.org updated to: 0.35-a0-736-g4e0084fca1