00:23:42 03hellmonk02 07[targepractice] * 0.35-a0-1246-g069b2e2b6e: fix debug 10(77 seconds ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/069b2e2b6e64 00:23:42 Branch pull/5501 updated to be equal with targepractice: 13https://github.com/crawl/crawl/pull/5501 01:02:23 03WizardIke02 07* 0.35-a0-1263-gc2a4187224: Fix out of bounds index in crawl.bane_name 10(5 minutes ago, 1 file, 5+ 1-) 13https://github.com/crawl/crawl/commit/c2a4187224ec 01:03:38 <11O​dds> Hmmm. There's a issue with how we do the randomness for passive mapping that could help with seed recovery. But I'm not sure how to fix it without doing a one-time reroll of the tiles revealed by passive mapping in ongoing games. Probably we just live with that? 01:11:00 03WizardIke02 07* 0.35-a0-1264-gca241dede9: Fix out of bounds index in crawl.bane_description 10(3 minutes ago, 1 file, 3+ 0-) 13https://github.com/crawl/crawl/commit/ca241dede995 01:14:15 03WizardIke02 07* 0.35-a0-1265-g90b9d4ed21: Remove duplicate include of state.h from l-crawl.cc 10(2 minutes ago, 1 file, 0+ 1-) 13https://github.com/crawl/crawl/commit/90b9d4ed2129 01:15:44 <04d​racoomega> That seems completely inconsequential to me 01:26:06 03CrawlOdds02 07* 0.35-a0-1266-g5f11247129: Run RC Lua on a non-seeded RNG 10(5 days ago, 3 files, 72+ 18-) 13https://github.com/crawl/crawl/commit/5f112471297f 01:26:06 03CrawlOdds02 07* 0.35-a0-1267-ge6b51bfc9a: Rework randomness in tile difficulties 10(27 minutes ago, 1 file, 12+ 9-) 13https://github.com/crawl/crawl/commit/e6b51bfc9a4d 03:31:59 <07w​izardike> @Odds This doesn't actually stop sometime from recovering the seed as the output of the subgenerator could be used to find its initial state which then gives you the seed. The quality of the randomness from the subgenerator is likely a lot better than from the hash based method, so this would likely make it a lot harder. However, you get the whole seed rather than just 31 bits of it now. I don't think the player gets output from 03:31:59 the subgenerator of high enough quality to reasonably reconstruct its initial state, but I'm not 100% sure on this. 03:34:41 <11O​dds> Agreed - but I think that's now like many other ways of getting the seed, where you are just trying to determine it from gameplay randomness (which as I understand it is hard because you get very small amounts of the entropy being generated). The special thing here was that you could attack half the seed, improving attack times by a factor of a billion 03:35:22 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-5261-gd9800d219b 03:35:45 <11O​dds> "I don't think the player gets output from the subgenerator of high enough quality to reasonably reconstruct its initial state, but I'm not 100% sure on this." - yeah, now we've removed the clua functions this is also what I think, including the part where I'm not 100% sure on this 🙂 03:37:03 <11O​dds> Hmmmm I guess passive mapping still is much faster than e.g. dungeon generation because the subgenerator starts in a simple state you can reconstruct. 03:41:39 <07w​izardike> The hash based method also required you to determine the seed from gameplay randomness though as the hash could spit out the same value for lots of different seeds 03:42:23 <07w​izardike> *although this randomness was probably of low quality making this easier 03:44:53 <11O​dds> My reading of the old thing was that enough mapping almost determined the 31 bits of the seed actually being used, do you think that's wrong? 03:45:13 <11O​dds> Like, there might be a few collisions 03:48:26 <11O​dds> Actually maybe not even that, looks like the hash we are using is invertible to me 04:02:16 <07w​izardike> Oh, yeah you're right. We were using murmurhash2 which would have been trivially reversable in this case as the known part of the data being hashed is at the end. The fact that we never give the result of the hash directly to the player would have helped, but this was likely a very poor choose of hash function for what we were doing 07:08:27 New branch created: pull/5504 (1 commit) 13https://github.com/crawl/crawl/pull/5504 07:08:27 03SeijiOokami02 07https://github.com/crawl/crawl/pull/5504 * 0.35-a0-1268-gd480f5cd81: Fix line-join glitches in Russian tutorial 10(36 minutes ago, 2 files, 8+ 8-) 13https://github.com/crawl/crawl/commit/d480f5cd8152 14:07:00 03SeijiOokami02 {dolorous} 07* 0.35-a0-1268-gbbeaa5dcc0: Fix line-join glitches in Russian tutorial 10(8 hours ago, 2 files, 8+ 8-) 13https://github.com/crawl/crawl/commit/bbeaa5dcc028 15:43:11 Unstable branch on underhound.eu updated to: 0.35-a0-1268-gbbeaa5dcc0 (34) 22:37:13 Unstable branch on crawl.develz.org updated to: 0.35-a0-1268-gbbeaa5dcc0 (34) 23:03:08 Windows builds of master branch on crawl.develz.org updated to: 0.35-a0-1268-gbbeaa5dcc0 23:41:39 Unstable branch on cbro.berotato.org updated to: 0.35-a0-1268-gbbeaa5dcc0 (34) 23:58:50 Monster database of master branch on crawl.develz.org updated to: 0.35-a0-1268-gbbeaa5dcc0