00:16:10 <09gammafunk> @pleasingfungus got that auto_explain issue fixed, for some reason the problem was setting a min duration of 2000ms; setting it to 0 to log unconditionally fixed it 00:27:08 03dolorous02 07* 0.34-a0-1866-g388b990e91: Add more mythical animals to artefacts. 10(8 minutes ago, 1 file, 4+ 0-) 13https://github.com/crawl/crawl/commit/388b990e912e 00:32:59 Unstable branch on cbro.berotato.org updated to: 0.34-a0-1865-gfffb249c6f (34) 04:31:57 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-5261-gd9800d219b 05:30:37 Unstable branch on crawl.akrasiac.org updated to: 0.34-a0-1866-g388b990 (34) 08:46:02 03dolorous02 07* 0.34-a0-1867-g2d85d88bbc: Add another orcish name, and update the count. 10(11 minutes ago, 1 file, 3+ 1-) 13https://github.com/crawl/crawl/commit/2d85d88bbce2 08:46:02 03dolorous02 07* 0.34-a0-1868-gfab9b65169: Add more vice names to artefacts. 10(7 minutes ago, 1 file, 6+ 0-) 13https://github.com/crawl/crawl/commit/fab9b651691e 10:22:35 <06pleasingfungus> very odd 11:10:46 New branch created: pull/4982 (1 commit) 13https://github.com/crawl/crawl/pull/4982 11:10:47 03CrawlOdds02 07https://github.com/crawl/crawl/pull/4982 * 0.34-a0-1867-g9b756b602f: Add messages for firing without anything quivered 10(9 hours ago, 4 files, 53+ 38-) 13https://github.com/crawl/crawl/commit/9b756b602fd8 11:52:39 03dolorous02 07* 0.34-a0-1869-g6a6fc1bf31: Add more orc names, and update the counts. 10(9 minutes ago, 1 file, 6+ 2-) 13https://github.com/crawl/crawl/commit/6a6fc1bf31b7 11:52:39 03dolorous02 07* 0.34-a0-1870-g55cff00ff5: Add another rare graffiti author. 10(8 minutes ago, 1 file, 2+ 0-) 13https://github.com/crawl/crawl/commit/55cff00ff5bb 12:43:02 <09gammafunk> it turned out that although the query I was running was taking 3-4 seconds, only 26ms of that was postgres. So there's some slowness happening downstream. Not to worry though, a more complicated query takes much longer postgres side and I have plenty to explain/analyze there 13:31:13 New branch created: pull/4983 (1 commit) 13https://github.com/crawl/crawl/pull/4983 13:31:14 03CrawlOdds02 07https://github.com/crawl/crawl/pull/4983 * 0.34-a0-1871-gcd4874107a: Fix preview of talisman attributes between integer skill levels. 10(5 minutes ago, 1 file, 9+ 6-) 13https://github.com/crawl/crawl/commit/cd4874107afe 15:52:45 03CrawlOdds02 {gammafunk} 07https://github.com/crawl/crawl/pull/4959 * 0.34-a0-1871-gf5bf96cdb1: Log piety tables 10(13 days ago, 15 files, 495+ 4-) 13https://github.com/crawl/crawl/commit/f5bf96cdb142 15:53:50 03CrawlOdds02 {gammafunk} 07* 0.34-a0-1871-g9b8c495655: Log piety tables 10(13 days ago, 15 files, 495+ 4-) 13https://github.com/crawl/crawl/commit/9b8c4956556a 15:54:55 03gammafunk02 07* 0.34-a0-1872-g4bf74e8489: Refactor out a small code block 10(2 hours ago, 1 file, 1+ 4-) 13https://github.com/crawl/crawl/commit/4bf74e84890e 15:54:55 03gammafunk02 07* 0.34-a0-1873-g37e25749c3: Dump piety info in wizard mode 10(10 minutes ago, 1 file, 8+ 1-) 13https://github.com/crawl/crawl/commit/37e25749c3c8 16:28:42 <11Odds> I've been debugging why monsters do loads of friendly fire when you are invisible. Turns out when you are invisible, monster tracer beams think anything within two squares of you is you. 16:31:17 similar to how it works for invisible monsters, where you occasionally see something within a square or two of its actual location 16:32:56 (I have to admit it was pretty amusing when an orc priest smote an invisible orc wizard that was attacking me) 16:33:22 <11Odds> Oh right yes! So they ignore the friendly fire because they think they've hit you first, I guess 16:34:51 <11Odds> (I think this could happen with a visible orc wizard too) 16:37:31 <04dracoomega> I was aware of this, but still surprised when someone reported monsters doing this with penetrating attacks. Whether they think you're there or not, I would expect them to still realize that their allies are also there. 16:37:52 04Build failed for 08master @ 37e25749 06https://github.com/crawl/crawl/actions/runs/20401175856 16:38:01 <04dracoomega> And they're typically quite reluctant to directly attack those 16:38:29 <04dracoomega> (One silliness that I did mean to address at some point was that a monster should at least not think the player is standing literally on top of another monster they can see) 16:40:05 <11Odds> Yeah there are two weirdnesses here: a) when the tracer considers the first test statue, it just thinks it's me and never processes it. b) it then stops processing anything because it's hit me so it can't possibly hit any more allies (?) 16:40:05 <11Odds> https://cdn.discordapp.com/attachments/747522859361894521/1452083160342073397/image.png?ex=69488553&is=694733d3&hm=759f7fd0134c032c575d5a6b09c3e9a30087cf74edd4d8813ca9841847786c7e& 16:41:34 Unstable branch on underhound.eu updated to: 0.34-a0-1873-g37e25749c3 (34) 16:43:25 <04dracoomega> Why would it think that? That is not how tracers for most things work. Are ranged attack tracers different than spell beams in that regard or something? Do they not understand that the javelin is penetrating? 16:46:57 <11Odds> b) is still confusing me... it does seem to know the javelin is penetrating. 16:56:40 <09gammafunk> this was a false alarm, something about docs deployment being in progress. presumably another CI task? 16:56:55 <09gammafunk> I guess if they trigger close enough together (via different jobs) you can have this happen 16:57:37 <09gammafunk> hrm, this seems to be related to me making those commits after the PR merge 16:57:39 <11Odds> Oh no... when it "finds" the player at the first test statue, it then fuzzes the targetting from that square to another. In this case in the walls. And then, unsurprisingly, things don't go well from there. 16:57:54 <09gammafunk> another sign that I should have found a way to push my commits in the same batch as that PR merge 16:58:06 <09gammafunk> maybe there's a gh pr command workflow to make that possible 17:05:31 <11Odds> (Actually I was misreading my logs, I think the tracer doesn't know the javelin is piercing; though I feel like there's other weirdness here with this fuzzing. Anyway I'll continue diving into this tomorrow,) 17:07:17 <06dolorous_84348> Quick note: For the record, when I said a few weeks ago that I'd have to take a step back from development soon, I'll have to actually do that in the next few days. In the meantime, good luck with everything, and I'll let you all know when things settle down again on my end; I can't really predict when that'll be at the moment. 17:20:27 <04dracoomega> Best of luck with whatever you're dealing with! 17:23:58 <08nicolae> agreed! good luck! 17:25:05 03DracoOmega02 07* 0.34-a0-1874-g727adfa5c0: Remove an unneeded parameter 10(28 hours ago, 3 files, 4+ 10-) 13https://github.com/crawl/crawl/commit/727adfa5c02e 17:25:05 03DracoOmega02 07* 0.34-a0-1875-gbb92ac17da: Fix xv of off-screen monsters sometimes showing wildly high damage numbers 10(26 hours ago, 5 files, 32+ 25-) 13https://github.com/crawl/crawl/commit/bb92ac17da00 17:25:05 03DracoOmega02 07* 0.34-a0-1876-gcc7c947500: Replace bound souls as a fineff instead of immediately 10(22 hours ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/cc7c94750077 17:25:05 03DracoOmega02 07* 0.34-a0-1877-g2f6f73a39c: Add a method to return the position of the last actor a beam hit 10(22 hours ago, 2 files, 44+ 0-) 13https://github.com/crawl/crawl/commit/2f6f73a39c6f 17:25:05 03DracoOmega02 07* 0.34-a0-1878-g78db3ca232: Tweak Bolt of Flesh behavior 10(22 hours ago, 2 files, 27+ 3-) 13https://github.com/crawl/crawl/commit/78db3ca2324c 17:25:05 03DracoOmega02 07* 0.34-a0-1879-ga887cd52da: Tweak Awaken Flesh behaviour 10(22 hours ago, 1 file, 42+ 39-) 13https://github.com/crawl/crawl/commit/a887cd52da4b 17:25:05 03DracoOmega02 07* 0.34-a0-1880-g4a4e20d81e: Don't wake monsters with BEAM_VISUAL 10(6 hours ago, 1 file, 0+ 7-) 13https://github.com/crawl/crawl/commit/4a4e20d81e7a 17:25:05 03DracoOmega02 07* 0.34-a0-1881-g9513272401: Make monsters a little less foolish when aiming cloud spells 10(6 hours ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/95132724011c 17:25:05 03DracoOmega02 07* 0.34-a0-1882-g22db19aefa: Remove impact damage from monster flame, noxious, and chaos cloud spells 10(6 hours ago, 3 files, 29+ 39-) 13https://github.com/crawl/crawl/commit/22db19aefa22 17:25:05 03DracoOmega02 07* 0.34-a0-1883-g173bb37f72: Rework BEAM_CHAOS functionality 10(4 hours ago, 1 file, 33+ 57-) 13https://github.com/crawl/crawl/commit/173bb37f725b 17:25:05 ... and 5 more commits 17:38:55 <08nicolae> one of these goddamn days i'm going to remember on the first try that the F in FTILE means "floor" and not "feature" 17:43:21 03nicolae02 07[nicolae1225-overflowing-cornucopia] * 0.34-a0-1636-g0ecf45e662: Add a few more god-specific overflow vaults 10(2 minutes ago, 1 file, 77+ 1-) 13https://github.com/crawl/crawl/commit/0ecf45e6621f 17:43:21 Branch pull/4956 updated to be equal with nicolae1225-overflowing-cornucopia: 13https://github.com/crawl/crawl/pull/4956 17:49:46 <08nicolae> anyway i probably won't be able to get to any updates on those vaults until after the holiday but i am open to comments 18:23:08 <09hellmonk> I can take a look later 18:23:30 <09hellmonk> Have been commissioned to make a few vaults myself 20:35:08 <09gammafunk> @o____0 after looking at analyze explain output I was seeing that a sequential index was just taking a very long time on ktyp even with the index, and we didn't have any parallelism. "NewSequell" was only 2vcpu and 4GB total ram, with only up to 2GB shared buffers. Initially it was doing a bitmap heap scan despite having an index, and after tweaking some things in accordance to us using an SSD, it was doing a sequential index scan, 20:35:09 but still slow. After some reading about different strategies (including https://medium.com/@jramcloud1/14-postgresql-17-performance-tuning-when-bitmap-scans-beat-index-seq-scans-1af21c97c04b) it seemed that using some parallelism would be helpful. I upsized sequell to 8vcpu and 16GB total ram, upped shared buffers to 4GB and tried it, but the plan seemingly wasn't parallel still 20:35:42 <09gammafunk> however when I lowered some relevant cost variables for starting a new process from 1000 to 0 (per recommendation), I got that result I showed you in the video 20:36:00 <09gammafunk> what's interesting is that explain analyze doesn't actually indicate that the query is now parallel 20:36:21 <09gammafunk> but it clearly did something in parallel after lowering that case 20:36:33 <09gammafunk> hence there are still many things I need to understand a bit more 20:37:15 <09gammafunk> as an aside, a query like !lg * name=gammafunk to get my last game actually only takes 26 ms in postgres but the full result takes close to 4 seconds to return in the run script 20:37:38 <09gammafunk> so there's some weird overhead happening too going from db result through the ruby/perl stack to output, or something 20:38:09 <09gammafunk> but one thing at a time: if normal queries aren't viable in postgres, we can't really have a bot 20:38:52 <09gammafunk> so I think I'm going to try open this up for queries on discord only and seeing what happens 20:39:44 <09gammafunk> I think we've still got a ways to go before this is truly "set up", and there's now a cost issue where this instance is $100/mo and the patreon can only cover about $50/mo of that 20:40:17 <09gammafunk> I don't know if we'll truly need this much cpu/ram once all is said and done, but I'll probably leave it that way for a while as things are tested and tuned 20:40:30 <09gammafunk> but long term, if that's how much is needed to run this, we do need to figure out a way to cover costs 20:42:59 <09gammafunk> at some point I could also email greensnark and ask for some advice on postgres tuning or whatnot, but for now I want to get this prototype set up for testing with a limited roll out but where it's continuously pulling records etc 23:35:41 Unstable branch on crawl.develz.org updated to: 0.34-a0-1888-g79e65ea6f0 (34) 23:39:24 03dolorous02 07* 0.34-a0-1889-g843a6e3170: Rename an obsoletely-named variable: mr -> wl. 10(14 minutes ago, 3 files, 7+ 7-) 13https://github.com/crawl/crawl/commit/843a6e31706f 23:59:01 Windows builds of master branch on crawl.develz.org updated to: 0.34-a0-1889-g843a6e3170