01:08:36 New branch created: pull/5214 (1 commit) 13https://github.com/crawl/crawl/pull/5214 01:08:36 03CrawlOdds02 07https://github.com/crawl/crawl/pull/5214 * 0.35-a0-291-g5575c309ce: Fix angry sphinxes timing out on recast (alenari) 10(11 minutes ago, 1 file, 6+ 8-) 13https://github.com/crawl/crawl/commit/5575c309ce33 03:01:05 03CrawlOdds02 07https://github.com/crawl/crawl/pull/5214 * 0.35-a0-291-g78732b0cb0: Fix angry sphinxes timing out on recast (alenari) 10(2 hours ago, 1 file, 8+ 7-) 13https://github.com/crawl/crawl/commit/78732b0cb073 03:35:44 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-5261-gd9800d219b 03:48:43 03CrawlOdds02 07https://github.com/crawl/crawl/pull/5214 * 0.35-a0-291-gd39ffc0c77: Fix angry sphinxes timing out on recast (alenari) 10(3 hours ago, 1 file, 8+ 7-) 13https://github.com/crawl/crawl/commit/d39ffc0c77be 09:50:32 <08nicolae> UPDATE: i slept in a bunch and will deal with the subvault flip issue TBD, for now i'm just going to try a tesselation of less-complicated shapes and hope that works 10:44:38 New branch created: pull/5215 (1 commit) 13https://github.com/crawl/crawl/pull/5215 10:44:38 03CrawlOdds02 07https://github.com/crawl/crawl/pull/5215 * 0.35-a0-291-gbbcad7cf3f: Correctly set targets on sculpted simulacra 10(2 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/bbcad7cf3f31 10:47:36 <11Odds> Perhaps on this fourth occasion that I've fixed an ally wandering off, I should think about whether there's a more systemic fix for this 11:16:08 03CrawlOdds02 07https://github.com/crawl/crawl/pull/5215 * 0.35-a0-291-g80cf9e72c0: Correctly set targets on sculpted simulacra 10(34 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/80cf9e72c0f7 13:24:06 New branch created: pull/5216 (1 commit) 13https://github.com/crawl/crawl/pull/5216 13:24:07 03CrawlOdds02 07https://github.com/crawl/crawl/pull/5216 * 0.35-a0-291-gc602e544f1: Implement a bias for exploring near stairs 10(3 weeks ago, 5 files, 110+ 32-) 13https://github.com/crawl/crawl/commit/c602e544f1ff 13:24:09 <11Odds> ^ Very much a draft, but if anyone is interesting in seeing how it feels I'd be interested in your opinion 16:30:24 The guidelines (https://github.com/crawl/crawl/blob/master/crawl-ref/docs/develop/contribution-process.md) for a pull request say to make a pull request in the main repository based off of the branch we made. Would I need my branch published for this? 16:31:55 is it on github? 16:32:05 if not, there's no way to make a pull request for it 16:35:03 I made a local repository clone like it said, made a new local branch, and that's where I'm at right now. The branch is on my github desktop app, but not on the repo 16:35:36 then it needs to be published to github in order to make a pull request, yes 16:36:03 unless someone is willing to accept a git patch set and put it up themselves, but then you need them again if you make any future changes 16:37:43 Oooooof, that makes this complicated then xP At least I think it would, considering I don't have permissions to upload the branch.... 16:40:36 normally when you fork a repo on github it goes into your own github account, and github tracks where it came from and can then make pull requests for the original repo from branches made on that fork 16:41:49 if you simply cloned it locally then there are ways to turn it into such a fork but they're a bit involved 16:43:56 Ooooooh, gotcha. I'm still pretty new to github contribution so that is definitely more of a lead than I have now. 16:43:57 Fun fact, it was two clicks to turn what i had into a fork lol, I just wasn't sure if that was the right move. Thank you so very much!! 16:45:04 interesting 16:47:57 New branch created: pull/5217 (1 commit) 13https://github.com/crawl/crawl/pull/5217 16:47:58 03Maddax F.02 07https://github.com/crawl/crawl/pull/5217 * 0.35-a0-291-g30e9b8af0f: fix(spells): Make binding sigil speed up players 10(21 minutes ago, 1 file, 4+ 1-) 13https://github.com/crawl/crawl/commit/30e9b8af0f29 16:59:32 <09hellmonk> would suggest modifying the description instead of doing that 20:25:54 <02Monkooky> Yeah I think every use case for self-swiftness post binding is rather tedious. 22:54:40 <02Monkooky> I'm somewhat dubious of this formula tbh ###### ..@<.. ###### Consider a corridor separated by a staircase, assuming no other staircases on the floor. To the best of my understanding, with a bias <= 10 you explore the entire left component before exploring anything near the stair. With a bias of 20, you oscillate left to right, exploring tiles that are within 3^i of the stairs on your ith oscillation With a bias of 10X, you 22:54:40 oscillate left to right, exploring tiles that are within ((X+1)/(X-1))^i of the stairs on your ith oscillation. I don't think your percentages check out here 22:57:05 <02Monkooky> Maybe this is the desired behaviour? I do think you're going to get a lot more than +100% explore time for 100 bias on bad level gens though 22:57:11 <11Odds> That's right, up to checking the exact exponents! 22:57:53 <11Odds> A bias of 100 is pretty silly behaviour indeed (empirically 100% is what I've seen, but that's on 5 levels and I didn't hit the old "infinite corridor" map) 22:58:45 <11Odds> Another interesting case is an empty plane, where I think the bias doesn't matter much and we just sort of spiral round 22:59:30 <02Monkooky> yeah I think this works generally well in empty spaces 23:00:43 <11Odds> The infinite corridor behaviour doesn’t seem at all silly to me at moderate biases 23:01:12 <11Odds> (My best guess is that 10 and 20 are pretty sensible settings) 23:01:20 <02Monkooky> distance is path distance, not just vector subtraction, right? 23:01:39 <11Odds> Yeah uses the whole travel path finding stuff 23:15:33 <02Monkooky> Yeah ok this is going to double back at silly times if you've got degenerate level gen- once you're on your 4thish oscillation with 20 bias and you're exploring 81 tiles out, there's not really much point in doubling back This isn't really a catastrophic fuckup or anything though, just kind of silly 23:15:51 <02Monkooky> (literal infinite corridor isn't needed to see degen nonsense, just a staircase in a chokepoint bisecting the level) 23:17:25 <11Odds> Yeah, long doublebacks do sometimes feel silly is my main observation 23:18:05 <02Monkooky> I'm not sure you can fix that without compromising how it handles distant path bifurcation though 23:18:56 <11Odds> I wonder if a formula that applies a sub linear function of stair distance works better 23:19:16 <11Odds> Though this one is as simple as it gets and mostly works imo 23:20:22 <11Odds> Very interested in how it feels in practice if you fancy checking it out 23:20:59 <02Monkooky> probably not playing today but I will get it downloaded and compiled 23:21:19 <02Monkooky> That was my initial feeling and why I went and did the math for how it currently behaves But this gets weird about how it handles split paths 23:22:52 <02Monkooky> ###.## ###.## #<.... ###### with a sublinear function, how far this fork in the path is from the stairs changes how I double back to explore both paths safely 23:23:14 <11Odds> Ah yes 23:24:07 <11Odds> One can imagine a thing that compared paths by how much they overlap to fix this. But one can’t imagine coding it. 23:27:24 <11Odds> I’ll carry on thinking if there’s some cleverer way. I am pretty certain I much prefer the current thing (at 10 or 20, day) to the normal autoexplore 23:46:46 Monster database of master branch on crawl.develz.org updated to: 0.35-a0-290-g6208fce011