00:03:39 <11O​dds> Yeah, not found these systematically yet but e.g. nicolae_lair_yak_garden always has a no_pool_fixup tag which from the code is intended only outside lair 03:34:53 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-5261-gd9800d219b 08:45:52 03CrawlOdds02 07* 0.35-a0-1309-g399d4016a6: Don't put random tags in the vaults cache 10(9 hours ago, 2 files, 9+ 6-) 13https://github.com/crawl/crawl/commit/399d4016a6ef 08:47:45 <11O​dds> This fixes the random cases which differ between builds; there appear to be many more that have dodgy logic because of the player's default depth at validation time; I'm just going to raise an issue for those as fixing them seems large and dubiously worthwhile 09:25:23 <09g​ammafunk> yeah, I wonder to what extent we can shield against this type of issue in the future, it seems fundamentally tied to doing random things during validation 09:25:44 <09g​ammafunk> Like, would it be possible to have dlua api functions that roll random results simply detect that we're validating and always return the same value? 09:26:09 <09g​ammafunk> you could imagine e.g. crawl.one_chance_in() doing something like this, for example 09:26:10 <11O​dds> In principle we could add a check that two rolls of the validation produce same result 09:26:27 <11O​dds> (Which actually is not currently true but not in ways that matter) 09:26:52 <09g​ammafunk> if we did a check like this, at what level would it operate 09:27:13 <09g​ammafunk> I was wondering about simply "turning off" randomness during validation for certain functions 09:27:13 <11O​dds> Deterministic but otherwise rolling randomly would be a bad answer I think, it’s still just wrong to add these tags 09:27:40 <11O​dds> Rebuild n times and check the .idx files are the same? 09:27:51 <09g​ammafunk> oh, like a CI step? 09:27:56 <11O​dds> Yeah 09:28:02 <09g​ammafunk> hrm, maybe 09:28:21 <09g​ammafunk> but I really worry about this having to put is_validating() check in so many locations 09:28:44 <09g​ammafunk> on the one hand you can say it's just something we have to be responsible for, I guess 09:29:01 <11O​dds> Another approach would be to define a set of tags that we ignore if added in validation 09:29:20 <11O​dds> (Some tags we need ofc for vault selection at least) 09:30:11 <11O​dds> But I think the tags that affect generation should probably always be added in the actual builder process? 09:31:40 <11O​dds> (I don’t know vaults stuff well enough to quickly know if tags actually split nicely like this) 09:35:05 <09g​ammafunk> well there are so many tags and vaults can just make up new ones whenever they want based on some notion. The latter I think are mostly the luniq_ style grouping tags that sets of vaults use to limit placement of that set 09:35:24 <09g​ammafunk> and the other fix you made in that commit wasn't even tag-related 09:44:51 <11O​dds> Pretty sure both were tag-related? 09:45:10 <11O​dds> But makes sense re: many tags 09:56:20 <09g​ammafunk> oh I see, so for the bazaar one, none of the map-related DES statements would matter in terms of seeds since they're not affecting what's in the vaults cache. 09:57:18 <09g​ammafunk> that would be just no_pool_fixup being added? 11:07:32 03hellmonk02 07* 0.35-a0-1310-g49955dfc25: Bugfix Five Virtues and adjust description 10(4 minutes ago, 2 files, 8+ 5-) 13https://github.com/crawl/crawl/commit/49955dfc25f2 11:42:49 FR: Make it a so that Call Merchant can't call a shop that is completely useless to your species i. e. armour shop for felid. DCSS is usually very good at avoiding that kinda stuff. You are still by no means guaranteed to get anything useful either away so I think there's no balance concerns. Thank you for listening to my Ted talk 11:43:26 (I'm surprised that it _can_ TBH) 15:42:16 <11O​dds> This is certainly the line that made me diagnose it as having a similar issue 15:43:41 <11O​dds> I think that’s the only thing that ends up cached there, though I’d have to check to be completely sure 15:43:48 Unstable branch on underhound.eu updated to: 0.35-a0-1310-g49955dfc25 (34) 15:49:07 <12g​e0ff> A felid could get an Assorted Antiques or a General Store with some armour and weapons in them, but you can't get an actual Armour Store. 16:59:24 ge0ff: thanks 22:30:46 Unstable branch on crawl.develz.org updated to: 0.35-a0-1310-g49955dfc25 (34) 22:41:50 Windows builds of master branch on crawl.develz.org updated to: 0.35-a0-1310-g49955dfc25 23:12:53 Unstable branch on cbro.berotato.org updated to: 0.35-a0-1310-g49955dfc25 (34) 23:45:29 Monster database of master branch on crawl.develz.org updated to: 0.35-a0-1310-g49955dfc25