00:10:05 <Cheibriados> 03advil02 07* 0.24-a0-118-g3d66cf4: Build for SSE2 on all systems, including 32bit targets 10(20 minutes ago, 1 file, 18+ 15-) 13https://github.com/crawl/crawl/commit/3d66cf40a65a
00:11:57 <gammafunk> I wonder if LTO would work now for cross-compiled builds
00:12:18 <gammafunk> I have it disabled for the debs too, since I ran into a build failure
00:12:28 <gammafunk> that was with an older debian though
00:12:46 <advil> it was causing some failures of the more inexplicable kind iirc
00:12:47 <gammafunk> I should test it, since I think we're using it for the OS X builds
00:12:57 <advil> yeah, we are
00:12:58 <gammafunk> oh, so it compiled but had runtime issues?
00:13:09 <advil> I can't really remember what was happening
00:13:19 <gammafunk> it's been a long time since I tried LTO with windows so my memory is hazy
00:13:40 <advil> I think a contributor or fork developer ran across this semi-recently
00:14:00 <advil> because I remember adding it to the docs not to use that
00:15:38 <advil> yeah, it was building but crashing when the player read a teleport scroll
00:16:41 <advil> rumflump reported it
00:16:52 <advil> no one ever figured out why it had that effect
00:18:06 <gammafunk> interesting
00:18:15 <gammafunk> maybe I should try a build and runs some qw tests with it
00:19:43 <gammafunk> well the contribs immediately fail to build it seems
00:19:52 <advil> heh
00:19:56 <gammafunk> cc1: error: -fno-fat-lto-objects are supported only with linker plugin
00:29:47 <Eksell> Unstable branch on CRAWL.XTAHUA.COM updated to: 0.24-a0-117-g53b5160 (34)
00:42:34 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.24-a0-118-g3d66cf40a6 (34)
00:47:57 -!- amalloy_ is now known as amalloy
00:54:53 <alexjurkiewicz> advil:does this mean if you use a system-supplied Lua you might still have seed instability?
00:55:00 <alexjurkiewicz> if it's compiled without sse2 support
00:58:08 <Cheibriados> New branch created: pull/989 (1 commit) 13https://github.com/crawl/crawl/pull/989
00:58:08 <Cheibriados> 03alexjurkiewicz02 07https://github.com/crawl/crawl/pull/989 * 0.24-a0-90-g289e282: Support profile-guided optimisation (PGO) 10(8 days ago, 2 files, 91+ 6-) 13https://github.com/crawl/crawl/commit/289e28273282
00:58:51 <advil> it's possible
01:02:49 <alexjurkiewicz> i guess draft pull requests still make Cheibriados notice
01:24:50 -!- yesno_ is now known as yesno
01:25:41 -!- amalloy is now known as amalloy_
01:28:34 <Gretell> Unstable branch on crawl.develz.org updated to: 0.24-a0-118-g3d66cf4 (34)
02:00:07 <Gretell> Windows builds of master branch on crawl.develz.org updated to: 0.24-a0-118-g3d66cf4
03:00:30 <Gretell> Monster database of master branch on crawl.develz.org updated to: 0.24-a0-118-g3d66cf4
03:26:44 <gammafunk> =nemelex MfAM
03:27:05 <Sequell> Defined keyword: nchoice => GhVM|TeEn|VpCj|VpAE|CeAE|DDIE|DEMo|FeWr|HaEE|GnHu|KoAK|FoSk|DDFE|OpAK|SpSk|MfAM
03:30:58 <Kelbell> Fork (bcrawl) on crawl.kelbi.org updated to: 0.23-a0-1729-gdd55c3d1e3
10:49:02 <advil> alexjurkiewicz: so I guess it may still be possible for a 32 bit linux setup to build with a system lua 5.1 that would include different rounding on conversion to int than everything else (now), but I can't figure out how common that is likely to be, it certainly doesn't happen on a current ubuntu install
10:51:41 <advil> most servers use contrib packages, right?
10:51:51 <advil> (all servers?)
10:53:27 <ebering> I think that trying to get seed stability over the divers universe of different system versions of dependencies is a quixotic task
10:55:06 <advil> well many things about this project are a bit quixotic
10:56:08 <advil> but I suspect this just isn't going to come up too much in 2019, so if that's so I might not worry about 32 bit linux
11:01:53 <advil> also, it would be nice (independent of seed stability) to have consistent rounding behavior across platforms
11:50:52 <advil> ok, I slogged through building crawl in 32 bit ubuntu in a docker container with lua from a deb, and can confirm that doing that will get "bad" rounding behavior
11:51:09 <advil> so I guess the question is whether anyone would ever do that
12:26:04 <gammafunk> !nchoice
12:26:06 <Sequell> Time for a new nchoice! It will appear shortly on the tournament website (if it hasn't yet). Type "=nemelex XXXX" to update !nchoice with the new combo, where XXXX should be replaced by the new combo.
12:26:11 <gammafunk> =nemelex BaAs
12:26:14 <Sequell> Defined keyword: nchoice => GhVM|TeEn|VpCj|VpAE|CeAE|DDIE|DEMo|FeWr|HaEE|GnHu|KoAK|FoSk|DDFE|OpAK|SpSk|MfAM|BaAs
12:31:08 <Henzell> Unstable branch on crawl.akrasiac.org updated to: 0.24-a0-118-g3d66cf4 (34)
12:32:06 <ebering> frog time
12:59:45 -!- Euph0ria is now known as EuphOria
13:19:23 <gammafunk> frogs? old news
13:19:28 <gammafunk> =nemelex TeEE
13:19:31 <Sequell> Defined keyword: nchoice => GhVM|TeEn|VpCj|VpAE|CeAE|DDIE|DEMo|FeWr|HaEE|GnHu|KoAK|FoSk|DDFE|OpAK|SpSk|MfAM|BaAs|TeEE
13:19:39 <gammafunk> tengus are the new cup of teee
15:33:32 -!- amalloy_ is now known as amalloy
16:27:22 <alexjurkiewicz> advil:good research! I can confirm CPO is not 32-bit, so I'm safe
18:22:31 <Lantell> Unstable branch on underhound.eu updated to: 0.24-a0-118-g3d66cf40a6 (34)
19:44:20 <alexjurkiewicz> where was that website showing the prototype lua documentation?
20:01:17 <Cheibriados> Crash on using stairs to D:15 13https://crawl.develz.org/mantis/view.php?id=11886 by Zukil
20:14:40 <gammafunk> alexjurkiewicz: http://csclub.uwaterloo.ca/~ebering/crawl/lua/
20:14:41 <gammafunk> ??lua
20:14:42 <Sequell> lua[1/1]: http://www.lua.org/manual/5.1/manual.html
20:15:00 <gammafunk> !learn add lua Draft documentation of crawl's lua api: http://csclub.uwaterloo.ca/~ebering/crawl/lua/
20:15:01 <Sequell> lua[2/2]: Draft documentation of crawl's lua api: http://csclub.uwaterloo.ca/~ebering/crawl/lua/
21:19:53 <advil> http://doc.dcss.io/
21:22:16 <advil> levels_seen = 16
21:22:21 <advil> well, that seems unusual
21:24:08 <advil> !lm Zukil
21:24:09 <Sequell> 1042. [2019-02-19 00:23:42] Zukil the Severer (L15 DDFi of Elyvilon) ASSERT((int)levels seen <= brdepth[branch]) in 'place-info.cc' at line 33 failed on turn 21295. (D:15)
21:24:17 <advil> !lm Zukil x=src
21:24:18 <Sequell> 1042. [2019-02-19 00:23:42] [src=cbro] Zukil the Severer (L15 DDFi of Elyvilon) ASSERT((int)levels seen <= brdepth[branch]) in 'place-info.cc' at line 33 failed on turn 21295. (D:15)
21:38:30 <advil> is it currently possible to get instantly shafted when taking a downstairs?
21:38:51 <advil> and if so, is it actually instant or does it take some time?
21:39:55 <ebering> I think it is not, I think it checks if you're standing on a shaft-ok square
21:40:10 <advil> ah ok
21:40:26 <advil> trying to figure out if there's some way I've missed to spend 0 turns on a level
21:40:34 <advil> hm, what if you take an escape hatch?
21:41:51 <ebering> yes, it is
21:42:36 <ebering> it's also possible to get banished after 0 turns if you hatch into los of a fast banisher who gets their banish on the first try
21:42:51 <ebering> (they have to be fast because most monsters don't get enough activation energy on the first entry)
21:42:56 <advil> nice
21:43:15 <advil> ok, well, the shaft version of this is probably the trigger, in combination with a change I made for pregen
21:43:21 <advil> lucky this hasn't come up more
21:56:37 <gammafunk> did it try to shaft them from d:15?
21:57:18 <gammafunk> !learn set lua[2] Documentation of crawl's lua API: http://doc.dcss.io/
21:57:19 <Sequell> lua[2/2]: Documentation of crawl's lua API: http://doc.dcss.io/
21:57:44 <gammafunk> dcss.io redirects to cdo, rats
21:57:49 <gammafunk> was hoping for a secret page
21:58:34 <alexjurkiewicz> man, i think the next project for dataification is monster definitions
21:58:58 <alexjurkiewicz> i was trying to expose all this info through the lua api and it's like writing a third facade for the data. There is already monster info, describe.cc the monster program Cheibriados runs
21:59:09 <alexjurkiewicz> fourth i guess
21:59:41 <ebering> well it's not that simple
22:00:04 <ebering> monsterinfo collects the data about a particular instance of a monster class that can be revealed to the player without considering it an iformation leak
22:00:11 <ebering> it includes static stuff from the monster class
22:00:48 <ebering> but also dynamic information such as (de)buffs, spell book as determined by observation, gear, attitude, and damage state
22:01:37 <advil> gammafunk: I'm not certain but I suspect what happened is that the player took an escape hatch to a new level (they had a couple of shafts in their milestones), and got shafted instantly. Some accounting in load_level interprets 0 turns as not having entered the level before, so the level got double-counted in the PlaceInfo for dungeon
22:01:58 <advil> really this is my fault at some level but
22:02:02 <advil> !flip PlaceInfo
22:02:03 <Sequell> (╯°□°)╯︵ °ɟuIǝɔɐlԀ
22:02:36 <gammafunk> and now you've reminded me how I have to rewrite the monstrosity that is !flip, so the day is ruined for both of us
22:02:39 <gammafunk> !cmd flip
22:02:40 <Sequell> Command: !flip => .echo (╯°□°)╯︵ $(map (fn (w) (apply concat (map (fn (c) (if (< $c T) (lookup $c (flipcharhash1)) (lookup $c (flipcharhash2)))) (reverse (split "" $w))))) (join " " (reverse (split " " ${*:-┻━┻}))))
22:03:22 <gammafunk> I think I may have messed with placeinfo once for the xp table
22:03:25 <gammafunk> maybe that's not the same data
22:03:31 <advil> it's related
22:03:53 <alexjurkiewicz> ebering: yeah it would be nice to reveal the dynamic data too. I don't see a way around that
22:04:31 <alexjurkiewicz> but at least I don't want to reimplement "monster base damage" and "monster can have a weapon" for the nth time in the codebase
22:04:45 <gammafunk> well you can also just start with monster class data
22:04:48 <gammafunk> not the dynamic stuff
22:04:56 <gammafunk> you know, what you can see with ?/
22:05:03 <gammafunk> ?/M I mean
22:05:04 <Sequell> No matches.
22:05:07 <gammafunk> ty Sequell
22:05:39 <gammafunk> I guess the dynamic stuff is the most interesting thing for lua writes
22:05:45 <gammafunk> isn't some of that already exposed though?
22:07:52 <alexjurkiewicz> actually it's the static stuff that interests me the most. Stuff like knowing the base damage, speed, hitpoints, and spells would let qw start to construct a more dynamic understanding of monster threat than the current hardcoded list
22:08:11 <ebering> yeah there's the monster.info class type
22:08:18 <ebering> (ok metatable type if you're a luahead)
22:08:46 <gammafunk> I like "luahead" sounds like the basis of a good clan name
22:09:40 <gammafunk> clan members: luamind LocalGlobal metatable G_
22:09:48 <gammafunk> or is that _G
22:10:00 <ebering> _G
22:12:04 <alexjurkiewicz> yeah I was writing up monster.info:base_damage() when I realised how much duplication would occur
22:18:57 <advil> haha, speaking of ways of being on a level for 0 turns, it's super easy to trigger this placeinfo corruption bug in wizmode
22:20:29 <advil> which is convenient for testing the fixup
22:33:05 <Cheibriados> 03advil02 07* 0.24-a0-119-g107517c: Prevent a PlaceInfo miscalculation and fix the resulting crash 10(6 minutes ago, 1 file, 17+ 1-) 13https://github.com/crawl/crawl/commit/107517c6e49a
22:34:09 <advil> if no one objects, after travis runs I'm going to cherry-pick that to the stable branch
22:35:47 <advil> haha oops
22:37:09 <Cheibriados> 03advil02 07* 0.24-a0-120-g429d40e: Set the branch depth correctly in 107517c6e49a 10(52 seconds ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/429d40e2d63a
22:37:35 <advil> if no one objects and I don't find any more disastrous errors in the fix, that is
22:39:16 <gammafunk> we say we have armel debian packages but that is a lie
22:41:18 <gammafunk> advil: maybe you're asking if anyone has technical objections, but in terms of crash fixes it's never a problem to cherry pick to stable
22:42:08 <gammafunk> sadly said fixes won't propogate to all servers like cwz and lld due to how those update
22:42:19 <advil> I've been being more conservative about changing stable this time around, because of seeding (not that this interacts with seeding)
22:43:00 <gammafunk> oh
22:43:13 <gammafunk> hrm, I wonder if that changes the general way we handle cherry picking to stable
22:43:26 <gammafunk> sounds like we can't just gradually cherry-pick fixes to stable unless we make a release?
22:43:35 <advil> yeah, ideally we do it all in one go
22:43:52 <advil> or we make yet another branch for the unstable version of stable
22:43:59 <advil> since that's what the repo needs, more branches
22:44:19 <gammafunk> well 98% or more of said "early" cherry picked commits to stable could certainly wait
22:44:35 <gammafunk> but there might be some important fixes and it could be pretty inconvenient to require a release for those
22:44:42 <gammafunk> due to the nature of said fix
22:44:53 <gammafunk> actually though
22:44:59 <gammafunk> this really is only relevant of online stable, no?
22:45:07 <gammafunk> offline stable users aren't even going to see this
22:45:16 <gammafunk> untill they update to a new release
22:45:36 <advil> yeah
22:45:47 <gammafunk> if that's true, then it'd possibly only become an issue if seeded runs became a thing on servers with automatic stable updates
22:45:57 <gammafunk> but cpo manually updates stable
22:46:15 <advil> online games do print a seed # in a death morgue now, though I could disable that
22:46:47 <advil> but for example it was kind of neat to be able to check ManMan's streak breaker to see what would've happened if he'd gone for the potion, etc
22:48:01 <gammafunk> Right, it feels like it's  more useful to be able to quickly cherry-pick important fixes to stable, but I guess that doesn't happen too much
22:48:01 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.24-a0-120-g429d40e2d6 (34)
22:48:01 <gammafunk> I'm trying to think of the last time we needed to fix something fast in stable
22:48:15 <advil> yeah, it's usually stuff like my commits above, which won't impact seeding so much
22:50:09 <advil> that kind of thing I don't have any objection to cherry-picking right away, would like to at least try for a while to keep seeding in sync. One options is to release more versions, usually we wait for a big batch
22:50:59 <advil> but for example I think we could do a 0.23.1 release next week
22:58:39 <gammafunk> yeah, we have tried to do point a first release soon after tourney, although often there is a delay
22:59:14 <gammafunk> I think the issue is that one person (myself) has to do a lot of builds and do all the uploads and updates and whatnot, so I'm not too excited to do frequent point releases
22:59:42 <gammafunk> it might be easier if we have more automated infrastructure
23:03:05 <alexjurkiewicz> fwiw, CPO stable only updates when you tag something on the stable branch (I do it by hand)
23:05:00 <gammafunk> oh
23:05:07 <gammafunk> good cpo trolling technique
23:18:03 <alexjurkiewicz> wtrunk is a branch, 0.23 is a release
23:24:16 <gammafunk> well there has to be a branch for 0.23 on your system as well, no?
23:24:35 <gammafunk> sounds like you're saying if we tag e.g. 0.23.1 cpo will automatically rebuild it?
23:24:47 <gammafunk> or you're saying you see the tag and then manually rebuild
23:48:26 <alexjurkiewicz> latter. But I would automate it if it would save me time
23:52:49 <gammafunk> yeah guess releases are too infrequent for that to matter much
23:58:51 <Cheibriados> 03advil02 07[stone_soup-0.23] * 0.23.0-1-gd1ef14e: Prevent a PlaceInfo miscalculation and fix the resulting crash 10(2 hours ago, 1 file, 17+ 1-) 13https://github.com/crawl/crawl/commit/d1ef14e61540
23:58:51 <Cheibriados> 03advil02 07[stone_soup-0.23] * 0.23.0-2-g0fe3ce5: Set the branch depth correctly in 107517c6e49a 10(83 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/0fe3ce5905b7