00:09:00 <12patrickm> ah, finally, i've done enough overflow vaults that each god has at least six of their own overflows (not counting shared ones or generic ones) 00:12:42 <12patrickm> i suppose i could try for 7 each but i should probably go to bed at some point 01:00:51 <10PleasingFungus> sleep good 01:34:49 Unstable branch on crawl.develz.org updated to: 0.27-a0-334-g52645efa06 (34) 01:55:13 Windows builds of master branch on crawl.develz.org updated to: 0.27-a0-334-g52645efa06 02:15:13 Unstable branch on cbro.berotato.org updated to: 0.27-a0-334-g52645efa06 (34) 02:53:47 Monster database of master branch on crawl.develz.org updated to: 0.27-a0-334-g52645efa06 03:30:53 Fork (bcrawl) on crawl.kelbi.org updated to: 0.23-a0-3917-gc4b44ba168 03:31:39 Fork (bcadrencrawl) on crawl.kelbi.org updated to: 0.03-131-g33da5a4ea5 04:17:00 -!- amalloy is now known as amalloy_ 06:05:15 exclusions on transporter entry portals are shown as being on "downstairs" in ctrl-o dungeon overview 13https://crawl.develz.org/mantis/view.php?id=12498 by hittemvvvhard 11:53:54 hello 11:59:28 Hello. 12:06:18 Unstable branch on crawl.akrasiac.org updated to: 0.27-a0-334-g52645ef (34) 12:59:41 -!- amalloy_ is now known as amalloy 13:11:39 wow, that's confusing 13:12:13 I'm getting a weird display problem on console, where my character occasionally gets displaced (often in place of a monster but occasionally around a wall or etc.) 13:13:04 it's actually where it "should" be so I end up e.g. using "h" to attack a monster visually on the "right", having to remember it was on the left 13:15:44 is it the @ symbol or just the cursor? 13:16:03 next time it happens I'll make a save 13:16:08 both 13:16:33 admittedly I'm playing with the "show species" option so it's actually a 9, conceivably that is somehow related 13:16:36 sounds like the sort of thing that might be fixed by a save and reload 13:16:54 advil: hey, before I go to Mantis, is it known that monster pathfinding is really expensive if you wake up W:1 because all the monsters in the disjoint bit pathfind all the way across it every turn? 13:17:17 geekosaur if you're on 0.26 or trunk you might try attacking with `v` (which gives you a targeter for all weapons now), in case that fixes the display 13:17:41 it's not constant, it seems to be triggered by a monster trying to attack 13:18:12 Pinkbeast probably in some broad sense of "known", not sure if there's already a report related to that 13:18:29 also what is W 13:18:31 and this is a rebuild as of a few minutes ago, to see if it's something that got fixed over the past couple of days 13:19:02 is it local then? 13:19:26 Tomb 13:20:24 yes 13:20:44 <10PleasingFungus> Womb 13:21:43 Hang on, it is still W in, say, shift-G W 3, you haven't changed it... 13:22:29 advil: I was looking at your commit ca9b790695f0bee32fb71c69 which fixes some of the symptoms 13:24:37 I can only remember like 3 of the branch abbreviations at any given time 13:25:03 C'mon, they're always coming up in pub quizzes 13:25:08 that commit that prevented a more in &~ has been killing me 13:26:12 fx: crosses second thing off list of weird vanilla behaviour to ask abotu 13:27:48 there are many similar effects like the one in ca9b790695 where there is a surprising call chain to check some monster properties (often has involved holiness) and gets some extra polynomials in the runtime, they only get optimized when they come up in particular cases 13:28:03 are you actually replicating this in regular dcss? 13:28:48 there was a bunch of world_reacts optimization that I did in 2019 13:29:04 of which that commit is one example 13:32:18 94313f71df4, 166592d65c2, 282e4f85318b2, d1d2e81aba 13:33:15 advil: In regular DCSS profiling shows that in this "woke up W:1" case monster pathfinding is eating most of the CPU cycles and on, say, V:5 it's not. I don't know a) if that's still enough CPU cycles to be ever problematic for servers and b) if there's anything that can be _done_ about it (if we couldn't path to where we're going at all and we're a long way from the player, wait a bit?) 13:33:16 you should also be aware that debug build timing is also not necessarily representative because the store validity checks are very heavy and run a lot (and are triggered by any of these call chain cases that uses one of the store classes at the bottom) 13:33:42 ime when profiling the problem is not usually the pathfinding itself 13:34:32 but some subcall that is accidentally called n^m times to get the same info on every pathfinding case 13:35:22 Mmm, but that's very much exacerbated by having to do a lot _more_ pathfinding 13:35:45 (arguably then doing less of it would make it harder to spot the problems you were fixing in 2019, yes :-/ ) 13:36:54 let me put it this way: every time I have started optimizing from the top of a call chain like this (and I have done it), that has been a mistake 13:37:01 display problems with multiple monsters 13https://crawl.develz.org/mantis/view.php?id=12499 by geekosaur 13:37:48 if the effect of more pathfinding is noticeable it is simply because the code it needs to do it is very, very unoptimized, but in a way that often doesn't matter 13:38:06 fwiw it was still broken on reloading the game after reporting the above 13:38:19 ok, well that should make it easier to replicate at least 13:38:25 that is, I had to move left to attack one of the monsters shown on the right 13:38:50 Hm. I'll go over the rest of your 2019 commits, then, and not report this since even if a) is true (and maybe it's not) this isn't the thing to fix. Thank you. 13:39:50 geekosaur it looks ok to me on loading that game, any chance you can also attach a screenshot to illustrate? 13:40:49 try moving to the left, which appears to be an open space but has an adder in it 13:41:02 it *looks* right, but it's not 13:41:07 yeah, I'm seeing something different I think 13:41:14 interesting :( 13:41:14 here's what I get on load: https://www.dropbox.com/s/zs8evudwoc1lnl5/Screen%20Shot%202021-02-13%20at%201.40.53%20PM.png?dl=0 13:41:46 advil: Thanks again for your help, especially since it doesn't seem to be a vanilla issue at all. 13:42:00 sigh, that looks like you're seeing it as it actually is instead of my display bug 13:42:25 guess I go rooting through my config and see if something with e.g. my replacement characters is sometimes causing problems 13:42:40 well, a screenshot would be useful with the report 13:43:36 Pinkbeast well I wouldn't be surprised if this case is not perfectly ok in regular dcss 13:43:59 Mmm, but I take your point that the top level isn't where to start. 13:44:27 W:1 is really good for exacerbating problems here because of the shape of the disjoint area - easy to wake everything up and a long way for them to walk to the other side 13:44:55 yes, that's why I've worked with places like V:5 and megazigs before also 13:45:04 I recommend actually going at it with a sampling profiler if you want to fix it 13:45:33 gotta redownload that save and then install a screenshotter. it's still reproducing here 13:46:36 Everything I know about profiling anything that isn't Perl, I learned this afternoon. :-/ 13:47:19 screenshot uploaded. I have a lot of replacement glyphs :) 13:47:41 but if it were that I'd expect it to (a) reproduce a lot more often (b) not sometimes do things like skip walls 13:47:58 have to repro one of those too, I guess, if I can 13:48:13 geekosaur thx...maybe upload your rc too given that? 13:49:23 attached now 13:49:58 Pinkbeast my experience is that mostly you just find some way to make the problem action repeat itself, start recording in the profiler, repeat the thing for 10-20 seconds, stop recording, invert call tree, and at least some of the problems pop right out (some require more investigation). I don't know much about the state of profiling on linux though, I use OS X Instruments which is very nice 13:51:06 ge 13:51:13 geekosaur It does replicate with your rc! 13:51:24 siiiiigh 13:51:54 (and I just repro'd it again, an adder "jumped around me") 13:52:10 I doubt it's the glyph overrides, like you said probably the player glyph change is the place to start, I will look into it a bit later 13:52:20 `v` at least places the cusor correctly btw 13:52:34 Once I'd worked out how to do it at all gprof pretty immediately showed most of the time was in Crawl hash tables and that led me to the mentioned commit, so... probably pretty useful if you aren't learning it all from scratch :-) 13:53:03 yeah for the hash table stuff, you can manually disable some of it if you want to stay on a debug build, or build with `make profile` 13:53:23 in general `make profile` gives much more realistic profiling info than make debug 13:53:40 the hash table stuff should generally be fully #ifdefed out on a regular build 13:53:57 I've considered disabling it on a debug build too, I'm not sure what it's really accomplishing in 2021 13:55:10 Oh, should have said - I wasn't profiling a debug build, it seemed like that was a bad plan. 13:55:25 03PleasingFungus02 07* 0.27-a0-335-gdc30354: Rewrite lightning scales' description 10(28 minutes ago, 1 file, 3+ 3-) 13https://github.com/crawl/crawl/commit/dc303548b972 13:55:25 03PleasingFungus02 07* 0.27-a0-336-ge8094ad: Remove boots of running 10(13 minutes ago, 15 files, 33+ 57-) 13https://github.com/crawl/crawl/commit/e8094add7c98 13:56:03 ok 13:56:03 well, if you see `assert_validity()`, that is the red herring 13:56:03 may still happen on a debug-lite build or something 14:11:49 04Build failed for 08master @ e8094add 06https://github.com/crawl/crawl/actions/runs/564312014 14:17:34 Unstable branch on crawl.kelbi.org updated to: 0.27-a0-336-ge8094add7c (34) 14:19:28 03PleasingFungus02 07* 0.27-a0-337-gc8f691b: Buff regen properties (Lici) 10(4 minutes ago, 3 files, 6+ 6-) 13https://github.com/crawl/crawl/commit/c8f691b2cc77 14:21:42 <12patrickm> i did, with the benefit of my new sleep powers i fixed up a couple minivaults that i felt were a little lackluster 14:21:51 <12patrickm> now they are fully lustrous 14:23:30 <10PleasingFungus> oo 14:23:31 <10PleasingFungus> shiny 14:23:45 <12patrickm> i already feel my old skills returning... it's like i've joined hepliaklqana but the ancestor i'm channeling is just me from like three years ago 14:24:01 <10PleasingFungus> what will you name yourself 14:24:04 <12patrickm> nicolae 14:24:10 <10PleasingFungus> bold 14:24:14 <10PleasingFungus> i like it 14:35:26 04Build failed for 08master @ c8f691b2 06https://github.com/crawl/crawl/actions/runs/564348643 14:36:20 blah. it's specifically the S glyph. but the way it's failing is still weird 14:47:03 Unstable branch on crawl.kelbi.org updated to: 0.27-a0-337-gc8f691b2cc (34) 14:48:02 Why was the Assassin change applied retroactively to already completed games/morgues? 14:49:32 Also sequell is bugged. 14:49:38 !lg * won BaAs 14:49:39 No games for * (won BaAs). 14:49:45 !lg * won As 14:49:46 3229. glosham the Heavyweight Champion (L27 TrBr of Cheibriados), escaped with the Orb and 15 runes on 2021-02-13 16:14:51, with 25519910 points after 58824 turns and 19:24:27. 15:46:04 <12patrickm> https://crawl.develz.org/mantis/view.php?id=12500 15:46:14 overflow vaults 13https://crawl.develz.org/mantis/view.php?id=12500 by nicolae 15:47:56 <12patrickm> oh, chei's already on it, nice 16:00:13 geekosaur I have a diagnosis for that bug but I have *no* idea how to fix it; for now my recommendation would be basically to avoid glyphs that come from right-to-left scripts 16:00:23 slightly more detail in the mantis response 16:00:27 so I saw 16:02:09 <10PleasingFungus> advil: ban rtl glyphs? 16:02:22 probably 16:02:58 not even sure how to identify them :D 16:03:08 at least for console 16:04:20 <12patrickm> how do you get right to left glyphs in console, anyway. do you have to specify them yourself 16:05:41 there's a .crawlrc attached to the issue 16:05:42 yeah, if you look at that bug report geekosaur gave their rc file and it has a whole bunch of stuff, in this case there's an arabic letter that looks like a snake 16:05:51 used for snakes 16:06:10 the rc just gives a unicode thing for it 16:06:52 I don't know why it doesn't always flip by my guess is that punctuation may still be LTR when mixed with a RTL glyph 16:07:27 but not numbers, so having `9` as that player triggers the bug but not `@` 16:07:42 *as the player 16:07:56 <12patrickm> weird 16:13:30 extremely weird 16:15:56 <12patrickm> tbh i'm pretty sure advanced unicode stuff is just some kind of sorcery 16:17:18 hm, I wonder if there's a way to keep a failing tag_upgrade build from cancelling the regular builds 16:23:53 03advil02 07* 0.27-a0-338-g9cfa79e: Fix an upgrade build issue for running removal 10(29 seconds ago, 1 file, 2+ 0-) 13https://github.com/crawl/crawl/commit/9cfa79e74d11 16:28:39 hm, looks also like the py2.7 CI checks are broken 16:28:50 in a very "why are you still using py27" way 16:32:17 someone who cares about things like flake8 and our somewhat over-elaborate requirements thing more than I do might need to fix that, my inclination would be to remove those tests 16:32:40 Unstable branch on crawl.kelbi.org updated to: 0.27-a0-338-g9cfa79e74d (34) 16:33:12 it looks to me like flake8 is simply refusing to install in py27 16:36:06 04Build failed for 08master @ 9cfa79e7 06https://github.com/crawl/crawl/actions/runs/564531697 16:37:07 maybe there's an easy fix 16:39:05 03advil02 07* 0.27-a0-339-gdf26f99: Fix more build issues 10(49 seconds ago, 2 files, 3+ 1-) 13https://github.com/crawl/crawl/commit/df26f997a615 16:47:02 Unstable branch on crawl.kelbi.org updated to: 0.27-a0-339-gdf26f997a6 (34) 16:47:11 <12ebering> RTL/LTR is a "terminal problem" 16:48:04 <12ebering> I put it in scare quotes because it's unclear what the correct behavior for mixed rtl/ltr for terminal emulators is and they seem to do different things 16:48:28 <12ebering> I use hebrew for L and W monsters without issue in iTerm2 e.g. 16:48:40 03advil02 07* 0.27-a0-340-g029c345: Actually update requirements files for the CI error 10(40 seconds ago, 2 files, 162+ 59-) 13https://github.com/crawl/crawl/commit/029c345baf4f 16:49:27 try playing a garg with the show_player_species option :) 16:49:42 yeah, this wasn't triggered by just RTL 16:51:03 though you are right that the bug doesn't seem to replicate on iTerm2 16:51:47 fun 16:51:48 so geekosaur if you are on mac you could just use that one (which is very nice actually) 16:51:56 linux 16:52:19 I kinda miss iTerm2 but can do without the rest of the Apple ecosystem 16:52:49 04Build failed for 08master @ df26f997 06https://github.com/crawl/crawl/actions/runs/564555973 16:55:04 <03wheals> i think i tried using ד for doors one time and putty just switched the text direction in the whole line after a door appeared 16:55:06 <03wheals> though ℸ works around that, if your font has it 16:57:14 augh why did you have to remove the first armour brand PleasingFungus 16:57:59 <10PleasingFungus> oh oops lol 16:58:01 <10PleasingFungus> what did i break 16:58:09 the tag_upgrade=yes builds 16:58:21 <10PleasingFungus> i don't know what that means 16:58:47 <12ebering> we made some baby steps towards upgrading to TAG_MAJOR = 35 while you were gone 16:58:54 <10PleasingFungus> oh i see 16:59:03 <12ebering> github ci checks that doing the tag upgrade works! 16:59:04 there's CI that removes all the TAG_MAJOR=34 stuff 16:59:14 <12ebering> "works" 17:01:03 I'm not actually sure how to correctly set the position of BRAND_ARM_FIRST in dc-item.txt, possibly there's no way without modifying the upgrade script 17:02:15 04Build failed for 08master @ 029c345b 06https://github.com/crawl/crawl/actions/runs/564571093 17:02:29 Unstable branch on crawl.kelbi.org updated to: 0.27-a0-340-g029c345baf (34) 17:03:38 <10PleasingFungus> advil: what if i create a new armour brand for boots and re-use running's enum, just to solve this 17:04:44 lol 17:06:39 hm maybe this supports commented out things for the future 17:07:09 ah yes it is already used for other cases like this 17:08:59 03advil02 07* 0.27-a0-341-g40fdf68: Continue fixing CI 10(34 seconds ago, 1 file, 4+ 1-) 13https://github.com/crawl/crawl/commit/40fdf68bc515 17:09:54 <10PleasingFungus> nice 17:16:06 03ebering02 07[ashenzaru] * 0.27-a0-331-gac44b50: fixup! fixup! Reforge the Chains I: Remove old Ash actives 10(31 hours ago, 1 file, 0+ 2-) 13https://github.com/crawl/crawl/commit/ac44b508318d 17:16:06 03ebering02 07[ashenzaru] * 0.27-a0-332-ge23de21: Reforge the Chains III: Curse piety 10(9 minutes ago, 11 files, 90+ 51-) 13https://github.com/crawl/crawl/commit/e23de210492b 17:16:34 Branch pull/1736 updated to be equal with ashenzaru: 13https://github.com/crawl/crawl/pull/1736 17:17:13 Unstable branch on crawl.kelbi.org updated to: 0.27-a0-341-g40fdf68bc5 (34) 17:27:13 <10PleasingFungus> curses! 17:34:14 04Build failed for 08ashenzaru @ e23de210 06https://github.com/crawl/crawl/actions/runs/564607927 17:37:29 <12ebering> @PleasingFungus I'm still working away but as I've hacked I refined my thoughts about the cursing/conduct 17:37:30 <12ebering> https://github.com/crawl/crawl/wiki/Ashenzari-brainstorm 17:37:32 <12ebering> updated this page 18:09:50 -!- Tiobot is now known as Guest19002 18:22:46 bye 18:23:55 Unstable branch on underhound.eu updated to: 0.27-a0-341-g40fdf68bc5 (34) 19:28:18 03PleasingFungus02 07* 0.27-a0-342-g5810d18: Make trolls more thin-skinned 10(9 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/5810d18f4bcc 19:28:18 03PleasingFungus02 07* 0.27-a0-343-gfcad680: Partially compensate VS regen 10(3 minutes ago, 1 file, 0+ 2-) 13https://github.com/crawl/crawl/commit/fcad680f6329 19:32:24 Unstable branch on crawl.kelbi.org updated to: 0.27-a0-343-gfcad680f63 (34) 21:02:03 <10PleasingFungus> i think it's fine for faith to work the same way as with ru (accelerate curse timer, freeish to take off at 6* [but maybe have an extra long timer before your next curse after you shatter chains]). 21:02:58 <10PleasingFungus> rest looks cool 21:19:01 <12patrickm> i feel like cursing an amulet of faith should do something special, but what 21:25:03 <10PleasingFungus> great q 22:07:57 <12patrickm> unique for swamp: William-o-the-Wisp, a unique will-o-the-wisp that summons more wills-o-the-wisp 22:15:08 <10PleasingFungus> a good name 22:43:36 -!- amalloy is now known as amalloy_ 22:47:53 -!- amalloy_ is now known as amalloy 23:03:14 @patrickm Thanks for the vault submissions, I'll take a look tomorrow 23:03:14 ....and reject all of them!!! 23:09:04 <12patrickm> nooooooooooo 23:51:31 <12patrickm> oh, hunh, you can place toxic bog just like any other feature 🤔