00:09:59 New branch created: pull/4277 (1 commit) 13https://github.com/crawl/crawl/pull/4277 00:10:00 03Isaac Clancy02 07https://github.com/crawl/crawl/pull/4277 * 0.33-a0-803-gd37a455533: Fix far viewing 10(10 minutes ago, 5 files, 95+ 30-) 13https://github.com/crawl/crawl/commit/d37a4555332b 00:28:02 03Isaac Clancy02 07https://github.com/crawl/crawl/pull/4277 * 0.33-a0-803-g11fb822da3: Fix far viewing 10(28 minutes ago, 5 files, 95+ 30-) 13https://github.com/crawl/crawl/commit/11fb822da381 00:31:14 03Isaac Clancy02 07https://github.com/crawl/crawl/pull/4277 * 0.33-a0-803-g353fe8153f: Fix far viewing 10(31 minutes ago, 5 files, 95+ 30-) 13https://github.com/crawl/crawl/commit/353fe8153fb4 00:42:56 03Isaac Clancy02 07https://github.com/crawl/crawl/pull/4277 * 0.33-a0-803-g138580c9b7: Fix far viewing 10(43 minutes ago, 5 files, 94+ 29-) 13https://github.com/crawl/crawl/commit/138580c9b7bb 00:47:39 Monster database of master branch on crawl.develz.org updated to: 0.33-a0-802-g62b7471ca3 01:13:24 <06m​umra> !tell nojaa At least with how throwing is currently, a missile shop would just be a bit boring for the majority of characters. After a possible throwing rework maybe some of the items could in fact be moved to gadget shops, and stones should be removed entirely tbh 01:13:24 <04C​erebot> docpaisley: OK, I'll let nojaa know. 01:20:26 <06m​umra> !tell nojaa (Also see nicolae_shop_firing_range, missiles alone aren't enough to carry the shop so it includes ranged weapons too) 01:20:28 <04C​erebot> docpaisley: OK, I'll let nojaa know. 01:22:27 <06m​umra> !tell nojaa I'm going to run objstat on the gadget shop branch and get some real numbers tbh, I don't see it imagine it making such an impact in Snake or Shoals for instance compared to the loot already in those branches) 01:22:28 <04C​erebot> docpaisley: OK, I'll let nojaa know. 04:17:56 <06m​umra> Running objstat, anyone know why i keep getting these messages about buggy bones files? 04:17:57 <06m​umra> https://cdn.discordapp.com/attachments/747522859361894521/1334482734940295239/image.png?ex=679cb165&is=679b5fe5&hm=abd869715d55ac6f0f2feee84b4ec86ad6456e65e96d63c09db08697d3db3bb4& 04:18:36 <06m​umra> (I tried deleting the whole bones directory, it just starts creating the files again and then giving this message anyway) 04:19:48 <06m​umra> Hmm, there's no way to know which condition in debug_check_ghost is rejecting 04:33:50 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-5261-gd9800d219b 05:27:55 <06m​umra> !tell nojaa Some quick stats from objstat comparing master to my gadget shops branch. Increase across a 3-rune game; wands - 5 avg, 17 max; evokers - 3 avg, 10 max. Actually less of an impact than I would have predicted and overall looks pretty reasonable. (Will put more in depth numbers in PR) 05:27:56 <04C​erebot> docpaisley: OK, I'll let nojaa know. 05:44:07 <06m​umra> slight bug spotted in objstat code as well 05:44:19 <06m​umra> c else if (is_shop_item(item)) { // XXX I don't think shops can place outside of vaults, but it would be // nice to find the item's true location, so we wouldn't have to // assume. -gammafunk in_vault = true; held_mons = false; in_shop = true; } 06:01:28 <06m​umra> Ok it's not exactly a bug but it's maybe a bit misleading to effectively include all shop items in the NumVault value as well (i didn't expect it anyway). seems more useful to remove in_vault = true here as see the number of floor items placed by vaults (if anything a new column could be added to distinguish random shops vs other vault specified shops but it still doesn't seem terribly useful) 06:36:39 -!- Changesite1 is now known as Changesite 08:10:32 <09g​ammafunk> I'm trying to remember why I felt it was so important to include shop items specifically as vault items. But I think treating them as floor items would be misleading as they're unavailable in a way that most floor items aren't 08:11:08 <09g​ammafunk> it's sort of a philosophical question and a bit of a question as to how you want to use objstat results 08:11:37 <09g​ammafunk> hrm, I should look at the surrounding code actually 08:13:20 <06m​umra> The random placement also uses a vault to place it, so yes all shops are vault placed -- so it sort of seems unneccessary to include it in NumVault 08:14:17 <09g​ammafunk> I suppose you could exclude them from the vault count and only have them be in the shop count 08:15:10 <09g​ammafunk> ah and there's not special "floor item" designation is there 08:15:37 <09g​ammafunk> it's just the overall count and there are the three subcategories for items that are monster/vault/shop 08:15:53 <09g​ammafunk> so I guess what you originally recommended does make sense to me; they'd still be in the overall count 08:16:05 <06m​umra> Unless as I say you checked which vault, if it's serial_shops then consider it part of "dungeon generation" rather than "vault placement", You'd still need an extra column to not lose data in one of the metrics 08:16:06 <09g​ammafunk> I had it in my head that we had a "floor item" category specifically, but we don't 08:17:06 <09g​ammafunk> I'll have to think about it a bit more later, but I could see the accounting logic changing 08:19:05 <06m​umra> you can have NumVault, NumShop, NumVaultShop. NumShop is those placed by serial_shops, NumVaultShop is those placed by anything else. NumVault is floor vault items. So total vault items is then NumVault+NumVaultShop, total shop items is NumShop+NumVaultShop 08:26:28 <06m​umra> i'm trying to work out if the distinction would even show me anything useful in what i'm looking at 🙂 09:03:22 <09g​ammafunk> Is serial shops truly the only "dungeon generation" vault? 09:03:37 <09g​ammafunk> I vaguely recall there being others 11:07:19 <06m​umra> it seems so. the ones that follow it (in shops.des) are subvaults to place different base types of shop, for use by "proper" vaults. everything else is nicolae craziness high quality custom shop vaults 11:08:37 <06m​umra> hmm, maybe i'm misunderstanding vault placement actually 11:34:10 <06m​umra> ah ... now i understand (serves me right for glancing over things a bit too quickly) serial_shops is deciding how many shops to place and what branches they are placed in. it can place any of the minivaults tagged with shop. these almost all have allow_dup, place a singular any shop, and have no DEPTH placement of their own, so should only get placed via serial_shops. there is as far as i can see only one exception here, 11:34:11 nicolae_shop_wares_on_display has the shop tag but also has a DEPTH line, and places a general shop instead. so i think it's possible for that to get placed in its own right, not just through the serial_shops logic; and also its DEPTH is !D:1-3 so i'm guessing it's possible for it to place anywhere, even the branches that don't normally get random shops? then, there are lots of "themed" shop vaults that don't have the shop tag and may have custom shops 11:34:12 or even multiple ones. and then we have more non-tagged shops in other .des files e.g. branch ends so if there's a distinction to be made, maybe we should be looking for the shop tag in the vault that placed it (since actually serial_shops will never be the local vault) 11:58:06 <08n​icolae> come to think of it, none of the vaults tagged shop are also tagged unrand so i think they can generate on their own without being placed by serial_shop, i think there might be a default-depth statement in shops.des somewhere 11:58:52 <08n​icolae> hm, weird. it might just default to the depth of serial_shop because it's the most recently defined vault with a DEPTH declaration 11:58:55 <08n​icolae> however i gotta go for a sec 12:08:43 <06m​umra> the objstat feature count for shop seems to suggest that's not the case otherwise we should be seeing way more shops overall 12:09:09 <06m​umra> in fact the shop count seems to be about half what i'd expect for most of D given a CHANCE of 15% 12:09:56 <06m​umra> wait, i'm wrong again 😭 12:11:20 <06m​umra> https://cdn.discordapp.com/attachments/747522859361894521/1334601864548978698/image.png?ex=679d2058&is=679bced8&hm=943c7dd39280d43de241b1ce13a3dc013bc30ff84f80239f820e56a080e1ac90& 12:11:39 <06m​umra> this is average number of shops placed on a given level, over 100 iterations 12:23:26 <06m​umra> serial_vaults should place 1.25 shops on average, when it places; and has CHANCE 15% in D, so 1.25 * 0.15 = 0.1875. from the numbers we're consistently getting a significantly higher average per floor, sometimes up to double ... 12:25:14 <06m​umra> (but i don't think we're getting too many shops overall. average shops in a 3 rune game is 19, although range is quite large: [10..33] 12:35:34 <06m​umra> nicolae_shop_wares_on_display hasn't placed at all yet in my mapstat runs 12:53:31 <08n​icolae> i wonder if that depth statement is screwing things up somehow 13:36:18 <06m​umra> the more i look at the mapstat results, the more confused i am. but yes, it's possible (it also has a WEIGHT ... and honestly vault placement is so byzantine it's hard to know what does or doesn't trigger random placement) 13:37:26 <06m​umra> mapstat has a section for "Random vaults available by dungeon level" and most of the shop tagged vaults are there, but not nicolae_shop_wares_on_display ... but serial_shops isn't there either, even though it demonstrably is placing a lot according to later on in mapstat 13:38:01 <06m​umra> it's also impossible to know from mapstat whether shops have been placed via serial_shops or they just placed by themselves 13:39:20 <06m​umra> wait i'm wrong again again ... the tagged ones aren't showing in random vaults available ... so maybe it's the tag that controls it 13:47:36 <06m​umra> @nicolae Confirmed, removing the DEPTH line allows it to place 13:48:21 <06m​umra> I think because there's an exclusion but no inclusion ... basically you're starting with nothing, and then excluding some floors. You need to first include some level ranges before you use an exclusion 13:50:16 <06m​umra> Some other vaults that might have this issue: nicolae_overflow_prove_your_dedication nicolae_spider_hilbert_zone_6 nicolae_spider_hilbert_zone_7 .... wtf is hilbert_zone.des 😂 14:06:05 <08n​icolae> just a bunch of jaggedy hallways for spider inspired by the hilbert curve 14:06:36 <08n​icolae> i can't fix those at work but if you have any quick fixes you can and want to push atm go for it imo 14:30:23 <06m​umra> well in conclusion i think - nicolae_shop_wares_on_display should probably have the shop tag removed and be moved to "themed shops", and have its DEPTH fixed. - for objstats, checking the shop tag is enough to provide a meaningful distinction between themed shops vs dungeon generation (but is this even useful, given that there are also a lot of branch end vaults that would count as the former, although you can also gauge them by looking 14:30:23 at the per-floor numbers) - something still feels a little off about the numbers, particularly how the averages vary so wildly, but maybe this would smooth out over more iterations 14:35:35 <09g​ammafunk> Is the official 0.32 objstat output helpful in this regard? It has 1000 iterations 14:35:42 <09g​ammafunk> ??objstat 14:35:42 <04C​erebot> objstat[1/2]: Spreadsheets for all released versions and trunk: https://drive.google.com/drive/folders/0B7VXhHzhWWb7S282VWhLVWRXbG8?resourcekey=0-USiftCbTwRlT9M7Uz5Fskg&usp=sharing/ ; See the README for details: https://docs.google.com/document/d/1D5mFqVi8ghz_nzvVmDUc3unx8VanVBWfgvZ8xCHaiJo/edit?usp=sharing 14:36:01 <09g​ammafunk> Not sure if you've seen this 14:43:34 <06m​umra> ah ... yeah, definitely smoothes out the numbers a bit 14:49:37 <06m​umra> seems to me overall that serial_shops is responsible for the majority of shops on most floors, random vaults making up between 0-30% of them depending on the branch/floor ... except for e.g. specific branch ends like Orc or V where obviously a lot of extra shops are placing 17:01:07 <09h​ellmonk> 19 shops is what, 170 shop items roughly? 17:01:37 <09h​ellmonk> some of which are not "real" purchases to consider ofc 17:40:41 <06m​umra> There's like a really high variance between min and max tho 17:40:43 <06m​umra> https://cdn.discordapp.com/attachments/747522859361894521/1334684754699878400/image.png?ex=679d6d8a&is=679c1c0a&hm=1cf7abdee63c89f786058fc58ecafd8422960ec0fab263bce8ee15f4718881e2& 17:41:06 <06m​umra> only 10 shops the whole game feels pretty rough ... 43 is wild tho 18:00:55 <09g​ammafunk> probably would be nice if objstat broke down the shop types 18:01:12 <09g​ammafunk> all shops do have a designated type, right? 18:08:15 <08n​icolae> yes 18:08:28 <08n​icolae> even custom ones use one of the "base types" as a, er, base 18:51:24 <09h​ellmonk> I do wonder if shops per level could use an adjustment. I know the idea is to make some floors feel themed around multiple shop vaults, but it does lead to these very long tail outlier games. Dunno if that's desirable for shops. 18:52:17 <09h​ellmonk> Then again, a decent number of shops are straight garbage (especially weapon and armour shops have a good chance to just roll junk) so it might be worth adjusting that too... 19:05:53 <04d​racoomega> None of the shop price adjustments got merged yet, did they? 19:19:31 <04d​racoomega> I keep falling down rabbit holes in the process of this vampire talisman stuff 19:19:42 <04d​racoomega> A new one every day, it feels like! 😛 19:20:11 <04d​racoomega> (This time to do with a bunch of monster tracer oddities, like refusing to cast dispel undead against player zombies if the player themselves isn't undead) 19:20:17 <04d​racoomega> Though the list is long 19:21:21 <04d​racoomega> Surveyed a big long list of issues and maybe-issues and then decided that I can probably push this off to after I merge the species (though I'd list to tackle at least some of them then) 19:25:06 <09h​ellmonk> No, it's still pending. I made a pr just so I don't forget it 19:27:05 <04d​racoomega> I guess you didn't feel finished with it yet? 19:34:39 <09h​ellmonk> Want some more testing and numbers feedback 21:37:39 Good evening