01:44:51 -!- amalloy_ is now known as amalloy
01:55:54 <Gretell> Windows builds of master branch on crawl.develz.org updated to: 0.21-a0-203-gb066a6e
02:53:28 <Gretell> Monster database of master branch on crawl.develz.org updated to: 0.21-a0-203-gb066a6e
03:11:50 <Rotatell> Unstable branch on crawl.beRotato.org updated to: 0.21-a0-203-gb066a6e (34)
04:26:50 <johnstein> !bug 11104
04:26:50 <Sequell> https://crawl.develz.org/mantis/view.php?id=11104
04:27:09 <johnstein> any suggestions on how to reproduce this?
04:33:34 <johnstein> separate question, where do I look to understand how monsters behave off-level?
04:57:45 -!- Telnaior_ is now known as Telnaior
05:02:05 -!- Telnaior_ is now known as Telnaior
05:27:17 -!- Telnaior_ is now known as Telnaior
07:09:01 -!- amalloy is now known as amalloy_
08:02:57 <gammafunk> johnstein: monsters don't act off-level. Off-level isn't even loaded into memory (except when examining things for travel/map)
08:03:53 <gammafunk> monsters do perform some wandering when you re-enter a level that simulates passage of time since you were last there
09:32:43 <|amethyst> !lg cybersaint2k ce 15 -log
09:32:44 <Sequell> 15. cybersaint2k, XL13 CeHu, T:17366: http://crawl.berotato.org/crawl/morgue/cybersaint2k/morgue-cybersaint2k-20170817-162213.txt
09:34:37 <|amethyst> any idea where the D:8 shadow dragon in that game would have come from?  In D they have depth 27, so even an 8 in a vault wouldn't spawn one, and the player was only on D:8 between about 800 and 1100 turns
09:36:50 <|amethyst> same player got a Sphinx in Orc:1 in their current game (it is a Xom game, but the sphinx "wanders into view" rather than being dumped on the player by Xom or polymorphed after being initially noticed)
09:47:10 <|amethyst> !tell dpeg I was looking at Orc maps and my wife, looking over my shoulder, asked of david_orc_1_choice "Why is there a Pac-Man ghost with 'NOO' written in it?"
09:47:10 <Sequell> |amethyst: OK, I'll let dpeg know.
10:12:20 <gammafunk> Does the noting actually discern between a monster that's never been seen and one that got polymorphed?
10:12:40 <gammafunk> If it got polyed and then shafted, for instance; I suppose that wouldn't be tied to the level in any way
10:12:50 <gammafunk> perhaps if the monster transited to a new level though
10:18:08 <elliptic> |amethyst: iirc the super out-of-depth spawns can start as early as 800 turns in, though they are extremely rare that early
10:18:34 <|amethyst> elliptic:  do they go that high?  (L 27 monster on level 8)
10:18:48 <|amethyst> gammafunk: re the sphinx:  https://www.reddit.com/r/dcss/comments/6uowvf/what_in_the_name_of_ood_is_going_on_in_trunk/
10:19:06 <|amethyst> gammafunk:  it says "a sphinx comes into view", which AFAIK means the player has never seen that monster before
10:19:20 <|amethyst> gammafunk:  so it would have to have been polymorphed out of LOS
10:19:34 <elliptic> I don't remember the formula but they go pretty high
10:20:08 <gammafunk> ??8
10:20:08 <Sequell> 8[1/2]: This is either a {statue} or a {golem}. It's either totally harmless or an electric golem.
10:20:10 <gammafunk> ??8[2
10:20:10 <Sequell> 8[2/2]: Is also a marker for extremely out of depth (depth*2+4 depth!) monsters in vaults; both traditional and valued by most devs. Good results: d:4 wolf spiders, d:5 centaur warriors, d:6 fire crabs, d:9 sphinxes, d:10 fire giants, d:11 liches.
10:20:21 <Cheibriados> New branch created: pull/599 (1 commit) 13https://github.com/crawl/crawl/pull/599
10:20:21 <Cheibriados> 03Sandman25DCSS02 {GitHub} 07https://github.com/crawl/crawl/pull/599 * 0.21-a0-204-g1749a14: Display range for spells in monster description 10(5 minutes ago, 1 file, 14+ 1-) 13https://github.com/crawl/crawl/commit/1749a149d0d6
10:20:28 <|amethyst> elliptic:  ah, I suppose that's possible
10:20:33 <gammafunk> oh those go past 8 don't they
10:21:00 <|amethyst> elliptic:  place.depth += random2avg(27, 2);
10:21:02 <gammafunk> elliptic: I have a prototype for that spawn counting code working, and am testing with qw; but I noticed that qw seems to "swing-rest"? and runs slower than I thought it might
10:21:16 <elliptic> swing-rest?
10:21:16 <gammafunk> is there a good way to run qw so that it's faster for testing purposes like this
10:21:32 <gammafunk> it went upstairs and did a bunch of "you swing at nothing"
10:21:50 <|amethyst> gammafunk:  did you set DELAYED = false already?
10:21:53 <elliptic> oh, that means there was an invis monster
10:22:08 <elliptic> probably an unseen horror downstairs
10:22:14 <gammafunk> oh ok, yeah and there had been a manticore, maybe it was Barb and there was display lag a bit
10:22:22 <gammafunk> or it just went by so fast and I didn't see that
10:22:31 <gammafunk> yeah I have DELAYED = false
10:22:48 <elliptic> how slow is it going? with DELAYED = false it should win in 10 minutes or some such
10:23:11 <|amethyst> you could try building with luajit :)
10:23:33 <gammafunk> it's about 10 minutes in, in orc:1
10:23:38 <gammafunk> seems to go through some long pauses
10:24:04 <elliptic> weird
10:24:07 <gammafunk> |amethyst: yeah I'm just trying to make sure I'm not doing anything weird that slows it down
10:24:20 <elliptic> is lua throttling on?
10:24:28 <elliptic> it usually isn't locally, but
10:24:49 <gammafunk> oh, this is a debug-lite build
10:24:55 <gammafunk> not sure if that matters
10:25:00 <|amethyst> oh
10:25:02 <|amethyst> no optimisation
10:25:02 <elliptic> that probably slows it, yeah
10:25:06 <gammafunk> aha, gotcha
10:25:17 <gammafunk> I should have a normal build on-hand for this then, thanks
10:25:36 <|amethyst> gammafunk:  it's possible to build with both debugging and optimisation
10:26:32 <gammafunk> doesn't that make debugging harder in various ways?
10:26:46 <|amethyst> yes, but less hard than without debugging at all
10:26:56 <|amethyst> make DEBUG=y   I think does it
10:27:28 <gammafunk> perhaps I'll just try a normal build and if I'm unlucky and my code has a crash, I can run it normally
10:27:44 <gammafunk> or maybe the backtrace will be sufficient anyhow so I won't need to
10:27:51 <|amethyst> NO_OPTIMIZE gets turned on by 'make debug' and 'make debug-lite'  but not by the otherwise equivalent   make FULLDEBUG=y DEBUG=y   and  make DEBUG=y
10:27:58 <|amethyst> (and by HURRY=y)
10:45:30 <gammafunk> resting seems pretty slow with qw sometimes
10:45:32 <gammafunk> wonder why that is
10:46:32 <gammafunk> maybe it's just all the lua processing going on that slows it down
10:58:39 <gammafunk> elliptic: table for a qw run, I think we just have to not count orb spawns for this to work: http://dpaste.com/24412GN
10:58:49 <gammafunk> something I hadn't thought of when implementing
10:59:19 <gammafunk> either that or they need their own category
10:59:50 <gammafunk> but I think we're not really interested in orb spawns so probably just omiting them would be good
10:59:55 <gammafunk> table is at the end of the morgue
11:00:40 <elliptic> is that the debug-lite build? I'm still curious whether a regular build is faster for you
11:01:02 <elliptic> yeah omitting orb spawns sounds good
11:02:04 <|amethyst> is it possible to have orb spawns on their own row?
11:02:45 <|amethyst> it's not relevant to the current question, but it does seem like the kind of thing that might be useful to have stats on
11:02:59 <gammafunk> elliptic: it was much faster with a regular build, yeah
11:03:06 <gammafunk> but there was still a strange slowness when resting
11:03:13 <gammafunk> I could see HP and aut ticking up slowly
11:03:15 <elliptic> it doesn't use 5 to rest
11:03:18 <gammafunk> aha
11:03:33 <|amethyst> gammafunk:  also, there are some alignment and spacing problems, but you probably already knew about those
11:03:42 <elliptic> I tried having it use 5 to rest and it didn't actually speed it up that much overall
11:03:43 <gammafunk> yeah, I need to fix that
11:03:51 <gammafunk> and yeah we could add orb spawns that way
11:03:51 <elliptic> and resting one turn at a time is simpler
11:04:18 <gammafunk> well, I'll time it when I run it the next time, but I think it was turning off optimization that slowed it down so much
11:04:26 <gammafunk> game did proceed very quickly with the new build
11:05:02 <elliptic> okay, good to know
11:05:27 <elliptic> looks like on most levels spawn xp was roughly 10%
11:05:50 <elliptic> though of course a lot of variance and seeing some stats on human games would be good too
11:06:41 <|amethyst> gammafunk:  hm, if you're looking for more stats, spawn XP (or count) / turns  might be useful?
11:07:16 <elliptic> but maybe something like removing later spawns and placing 10% extra monsters at start and having the extra monsters be awake would be good
11:07:41 <elliptic> though with awake monsters we want to make sure the player doesn't enter the level too near them
11:07:51 <elliptic> so that's a little tricky maybe
11:08:10 <gammafunk> certainly could ensure that for stairs
11:08:41 <gammafunk> Even doing that for new levels in terms of shaft/hatch destinations would be possible
11:09:11 <gammafunk> although I'm not sure about the current order for level gen
11:09:26 <gammafunk> I think you'd have to have that pass at a later stage from normal monster gen
11:09:53 <elliptic> another option would be to give monsters a very small chance of waking up each turn
11:11:55 <gammafunk> |amethyst: tricky part with spawn xp / turns is that turns are trucked per-branch (via PlaceInfo)
11:12:28 <gammafunk> let me paste my still pretty bad code for reference, but it would require rethinking how PlaceInfo works or some kind of merge or something
11:12:41 <gammafunk> and PlaceInfo is used in some context where it needs to work per-branch I believe, so that may not be simple
11:13:39 <Cheibriados> New branch created: xpinfo (1 commit) 13https://github.com/crawl/crawl/tree/xpinfo
11:13:39 <Cheibriados> 03gammafunk02 07[xpinfo] * 0.21-a0-194-g8c08ff6: First pass at xp info dump 10(43 minutes ago, 11 files, 332+ 37-) 13https://github.com/crawl/crawl/commit/8c08ff62f633
11:14:16 <gammafunk> it has not save compat at the moment
11:14:21 <gammafunk> s/not/no/
11:14:38 <gammafunk> also s/trucked/tracked/
11:17:38 <johnstein> !bug 10971
11:17:39 <Sequell> https://crawl.develz.org/mantis/view.php?id=10971
11:18:31 <johnstein> gammafunk: so this one is marked simple. I'm assuming that's because the solution is to just remove all statuses right when the monster shafts?
11:19:14 <|amethyst> gammafunk:  hm, how does the OOD timer work then?
11:19:49 <johnstein> and I suppose the HP should be set back to max? or leave the HP as it is?
11:20:11 <|amethyst> gammafunk:  ah, it's stored in the level itself
11:20:23 <|amethyst> gammafunk:  which I guess is a bit of a pain to access
11:20:34 <gammafunk> oh
11:20:52 <gammafunk> I assumed it was in placeinfo; I don't think access is necessarilly a problem, but I was more concerned the data simply wasn't there
11:21:43 <|amethyst> johnstein:  I'd think the solution would be to apply the same kind of enchantment timedowns as happens to monsters when you return to a level
11:22:04 <|amethyst> johnstein:  which AFAIK includes healing over time
11:23:16 <johnstein> oh ok. I think I get it. my training item is enchantment timedowns
11:24:23 <johnstein> so we handle monsters behaving offlevel with enchantment timedowns?
11:24:34 <johnstein> i.e. indirectly
11:25:26 <|amethyst> johnstein:  it's that "simulates passage of time since you were last there" that gf mentioned
11:25:36 <|amethyst> I'm trying to find it
11:26:29 <johnstein> ah I missed that second message that gf sent after the first one.
11:27:36 <|amethyst> johnstein:  update_level in timed-effects.cc
11:28:01 <|amethyst> johnstein:  you'd want to split out the stuff inside the loop into a function that can apply it to a particular monster
11:28:43 <|amethyst> johnstein:  then you need to do that from somewhere in place_transiting_monsters
11:28:55 <johnstein> thanks |amethyst
11:29:18 <|amethyst> johnstein:  *but* I'm not sure if the transit list knows what time the monster was added
11:29:27 <|amethyst> johnstein:  which you need so you can tell how much time to simulate
11:30:10 <johnstein> how does it work when leaving a level? wouldn't you need to set the time the monsters were added then?
11:30:35 <johnstein> (or maybe that might make more sense once I absorb the code)
11:31:10 <|amethyst> johnstein:  for the case that's handled now, you just need to know the time you left the level and the current time
11:31:19 <|amethyst> johnstein:  and the time you left the level is stored in the level itself
11:31:24 <johnstein> ah good point
11:31:49 <|amethyst> johnstein:  but the time the monster left the level would have to be stored in the 'follower' struct
11:31:53 <|amethyst> (mon-transit.h)
11:32:41 <johnstein> !source mon-transit.h
11:32:42 <Sequell> https://github.com/crawl/crawl/blob/master/crawl-ref/source/mon-transit.h
11:32:45 <|amethyst> basically, when a monster leaves the current level for another one, we put it in a separately marshalled list (called "the_lost_ones"), tracking what level it's destined for
11:33:47 <|amethyst> then, when you enter a level, place_transiting_monsters looks for monsters on the transit list that are destined for that level, removes them from the transit list, and places them on the level
11:34:13 <|amethyst> there is a separate list, that works the same way, for monsters following you across stairs
11:34:14 <johnstein> why are they called "followers"?
11:34:30 <johnstein> oh
11:34:30 <|amethyst> preemptive strike :)
11:34:35 <johnstein> so not allies. more like predators
11:34:38 <|amethyst> both
11:34:43 <gammafunk> they're worshiping you
11:34:51 <gammafunk> they express devotion by taking your HP
11:34:58 <johnstein> how sacred
11:35:16 <|amethyst> "follower" as in quite literally "following you across the stairs"
11:35:30 <|amethyst> no moral judgment either way
11:35:39 <johnstein> so monsters following you to a new level or pre-following you via a shaft are both considered the same kind of follower
11:36:04 <johnstein> ?/pre-follow
11:36:05 <Sequell> No matches.
11:38:20 <|amethyst> johnstein:  well, they use the same data structure, but they go onto different lists
11:39:59 <|amethyst> oh, no, sorry, I'm wrong
11:40:17 <|amethyst> it is the same list, they just have a flag set
11:40:21 <|amethyst> MF_TAKING_STAIRS
11:41:29 <|amethyst> place_followers handles the ones that are taking stairs (through _level_place_followers), and place_transiting_monsters handles the others
11:41:39 <|amethyst> (via _level_place_lost_monsters)
11:46:46 <johnstein> what are other ways to lose monsters besides shafts?
11:55:24 <|amethyst> hm banishment
11:56:20 <travis-ci> The build passed. (xpinfo - 8c08ff6 #8632 : gammafunk): https://travis-ci.org/crawl/crawl/builds/266297708
11:57:00 <|amethyst> johnstein:  looks like that is everything
11:58:11 <|amethyst> monster::set_transit() is the relevant method
11:58:30 <johnstein> probably not as important, thematically or practically, for abyssed monsters to get their statuses cleared since there's less chance to see them again in the same state and if you do, you can just blame abyss weirdness as the "reason".
11:58:44 <johnstein> and probably not noticed as much as shafted
11:59:21 <|amethyst> johnstein:  but probably it makes sense to handle it everywhere
11:59:31 <johnstein> I agree
11:59:58 <|amethyst> particularly since the relevant code would need to run when the monster is pulled *off* the transit list
12:00:14 <|amethyst> meaning you have lost the information about how it got on there in the first place
12:00:32 <johnstein> was just saying that I can see why it hasn't been noticed if these are the only two ways to see the issue and one way is the abyss where things are already goofy
12:00:37 <johnstein> agree
12:00:39 <|amethyst> yeah
12:00:43 <|amethyst> also, even with shafts
12:00:56 <|amethyst> it could take a while before you autoexplore to the monster
12:01:07 <|amethyst> by which time it has healed and the enchantments have timed out normally
12:01:19 <johnstein> neat little bug
12:01:38 <johnstein> feels a bit less simple than I was expecting based on the tag, but still doesn't sound that bad
12:03:47 <|amethyst> yeah, I'm not sure the tag was really appropriate there
12:04:16 <|amethyst> maybe whoever added it was thinking of your simple proposal of just healing the monster and clearing non-permanent enchantments
12:09:29 <|amethyst> BTW, not relevant to this, but recall works completely differently
12:15:20 <johnstein> if you have an ally who gets poisoned and gets shafted and you recall them 1000 turns later, would they suffer from this big and come back poisoned and the same HP? or is that what you mean by handles differently?
12:15:47 <|amethyst> yeah, time is not simulated for offlevel recall
12:16:41 <|amethyst> also
12:16:43 <johnstein> should it be?
12:16:57 <|amethyst> maybe, but don't worry about that now
12:17:04 <|amethyst> you don't want to open that can of worms
12:17:22 <johnstein> ah ok
12:17:27 <|amethyst> all the "unlinked item" warnings about orcs are *probably* because of offlevel recall
12:17:55 <|amethyst> also, even with a shaft
12:18:10 <|amethyst> the poison expires but doesn't actually hurt the monster
12:19:24 <|amethyst> timeout_enchantments doesn't actually apply any of the enchantment's continuing effects, it just decreases the duration and removes the enchantment if necessary
12:19:40 <|amethyst> also, I think this is one for @crawlcode:
12:19:40 <|amethyst> * @param levels XXX: sometimes the missing aut/10, sometimes aut/100
12:20:26 <|amethyst> I guess I should change that @param to \param for twitter purposes
12:21:32 <|amethyst> hm
12:21:42 <|amethyst> most of that header comment is somewhat obsolete
12:21:59 <|amethyst> the "current abuse" mentioned there is pre-DCSS, since the comment is from Brent Ross
12:22:07 <johnstein> ok. so for the original bug success looks like, poisoned 50% hp monster gets shafted. wait 1000 turns. go see it in the new level and confirm it's not poisoned and at 100% HP.
12:22:12 <|amethyst> not sure I would call 10 years "current"
12:22:17 <johnstein> heh
12:22:40 <|amethyst> johnstein:  that but also try it for fewer turns
12:22:48 <|amethyst> johnstein:  the monster should be in a state in between
12:23:00 <johnstein> if instead you waited precisely the amount of time the poison enchantment you would see no poison and still 50%
12:23:14 <|amethyst> no poison and higher than 50%
12:23:20 <|amethyst> since it still healed over time
12:23:27 <|amethyst> but maybe not to full
12:24:05 <|amethyst> would work best with a low HD monster with lots of HP
12:24:05 <johnstein> yes. that's actually what my brain was trying to say. that since it won't apply the poison dmg the hp state won't be quite accurate
12:24:18 <johnstein> but only inaccurate for a certain transition period
12:24:19 <|amethyst> since monster regen rate is based on HD
12:24:34 <johnstein> and players won't notice it
12:24:36 <|amethyst> so you could do a goblin HP:9999 or something
12:25:02 <johnstein> ah ok
12:25:37 <johnstein> is there some place that documents all the fancy ways you can create monsters and items?
12:25:43 <|amethyst> err, I guess that's probably lowercase hp
12:25:59 <johnstein> or is it always just the attribute:value?
12:26:08 <johnstein> comma or space separated?
12:26:17 <johnstein> for multiple attributes
12:26:21 <|amethyst> johnstein:  docs/develop/levels/syntax.txt, look for MONS: and ITEM:
12:26:34 <johnstein> thanks
12:26:34 <|amethyst> johnstein:  but there are probably still some undocumented things there too
12:26:45 <|amethyst> johnstein:  comma separated in general
12:26:54 <|amethyst> johnstein:  err
12:26:59 <|amethyst> johnstein:  space separated I mean
12:27:24 <|amethyst> johnstein:  but things get more complicated when you have things like a monster carrying multiple items
12:27:45 <|amethyst> still space separated tags, but then you need other characters to separate the monster from the item and the item from the next item
12:28:12 <|amethyst> mostly key:value but sometimes it's just a tag like   goblin always_corpse
12:29:14 <|amethyst> most tags can go anywhere, including before the monster name
12:29:40 <|amethyst> even between words:  att:friendly orc hd:8 warrior
12:29:58 <|amethyst> (please don't do that :) )
12:33:41 <johnstein> by golly it worked
12:45:32 <|amethyst> johnstein:  so now the real difficulty
12:45:34 <|amethyst> johnstein:  save compat
12:46:01 <|amethyst> johnstein:  or, rather, saving the transit list entry time, and then handling the lack of that in old saves
12:47:53 <|amethyst> (for the save compat, it would be fine to just set that time to 0 and let things time out/heal completely; but you do at least need the code to not try reading the saved time if the save is old enough)
12:57:09 <gammafunk> elliptic: Here's a table without any orb run spawns (the table alignment is messed up a little, but it should still be clear)
12:57:22 <gammafunk> I'm sure you'll appreciate qw's choice of weapon
12:57:56 <gammafunk> Would it be good to run qw as a human, perhaps just HuBe?
12:58:07 <gammafunk> I'm not sure what happened on Zot:5 with the 0 spawn situation
12:58:25 <|amethyst> for testing that you will need a save made with trunk crawl (without any your changes) with two poisoned hurt monsters on the transit list: one that has been on there a long enough time for the afflictions to go away, and one that just went on
12:58:43 <gammafunk> I did see what looked like a teleport at one point, but am a bit surprised that it would clear zot:5 without any spawns
12:59:02 <|amethyst> and back that save up, since you will be loading/playing several times
13:00:50 <|amethyst> to test whether saving works at all, save and re-load the game in between the monster getting shafted and the player entering the new level
13:01:12 <gammafunk> elliptic: er, forgot to give the url: http://dpaste.com/0QRVN2K
13:02:30 <|amethyst> probably with the same old + new monster setup, so you can check both possible failure modes (doesn't time anything out vs times out too much, as though the monster was shafted on turn 0)
13:02:38 <|amethyst> s/was s/were s/
13:10:19 <elliptic> gammafunk: 0 spawns on zot:5 is weird yes, I wonder if spawns don't happen if there are a ton of monsters on the level already or something?
13:10:53 <elliptic> or maybe it just isn't that unlikely
13:11:52 <elliptic> gammafunk: if it's not too hard to list the number of turns spent on each individual level alongside the spawn stuff, that might be interesting
13:12:21 <gammafunk> hrm, I'll look into it
13:14:08 <elliptic> I guess a bunch of other levels had 0 spawns killed also
13:14:34 <gammafunk> let me push my current code right quick, but I don't think I'm miscounting
13:17:00 <Cheibriados> 03gammafunk02 07[xpinfo] * 0.21-a0-195-gf8a6eaa: Fix (not quite) xp_by_level table alignment 10(2 minutes ago, 1 file, 5+ 5-) 13https://github.com/crawl/crawl/commit/f8a6eaa34449
13:17:00 <Cheibriados> 03gammafunk02 07[xpinfo] * 0.21-a0-196-g65211fe: Don't include Orb spawns in the spawn count of xp_by_level 10(38 seconds ago, 1 file, 3+ 2-) 13https://github.com/crawl/crawl/commit/65211fecd5a7
13:17:29 <gammafunk> I wonder though, is this handling ood monsters?
13:20:05 <gammafunk> yeah looks like it is
13:20:20 <gammafunk> it becoming an ood or super ood is just in the choice of random monster that happens later
13:43:02 <Cheibriados> Dithmenos Wrap-arounds Some Text 13https://crawl.develz.org/mantis/view.php?id=11188 by Pekkekke
14:14:47 -!- amalloy_ is now known as amalloy
14:45:31 -!- mumra1 is now known as mumra
14:52:21 <gammafunk> elliptic: qw HuBe that died in the zot:5 orb chamber: http://dpaste.com/072BDVH
14:52:43 <gammafunk> One thing I realized about that spawn XP/count is that this doesn't mean that spawns aren't generating, of course, only that they're not getting killed
14:53:13 <gammafunk> I could clean up the commits, add proper save compat, and then push this to trunk
14:53:25 <gammafunk> I guess we could just enable this table in default options?
14:53:37 <gammafunk> That's what I actually do right now
14:55:35 <gammafunk> hrm, wait
14:56:02 <gammafunk> this turn total
14:58:10 <gammafunk> doesn't seem to agree with the overall turncount at all, seems way inflated
15:28:06 <Cheibriados> 03alexjurkiewicz02 {MarvinPA} 07* 0.21-a0-204-ge2129ae: Gozag: don't offer useless haste/berserk in petition 10(5 weeks ago, 1 file, 11+ 0-) 13https://github.com/crawl/crawl/commit/e2129aebca54
15:36:47 <elliptic> gammafunk: hm, column F in the previous table (average non-inter-level travel turns spent in a level) also seems inflated
15:38:16 <elliptic> e.g. that game spent less than 8000 turns total in zot, but that table says 4620 per level
15:38:41 <elliptic> I don't know whether it is using turns or aut, but it shouldn't make that big a difference
15:39:08 <gammafunk> elliptic: there's some kind of fundamental problem with how turns are being counted for place-info purposes, as I'm learning, let me paste something
15:39:41 <gammafunk> http://dpaste.com/0GZ3D8F
15:39:47 <gammafunk> That's not actually qw, confusingly
15:39:51 <gammafunk> that's me playing normally on D:1
15:39:54 <gammafunk> quitting at the end
15:40:06 <gammafunk> not that the turns recorded are about 1/3 of the actual turns passed
15:40:23 <gammafunk> I think the call to _update_place_info() in main.cc
15:40:35 <gammafunk> is probably not correctly handling turns that happen under a delay
15:40:45 <gammafunk> also let me push my current code
15:43:07 <Cheibriados> 03gammafunk02 07[xpinfo] * 0.21-a0-197-g724e2f1: Add turn tracking to the xp_by_level dump table 10(2 minutes ago, 5 files, 29+ 10-) 13https://github.com/crawl/crawl/commit/724e2f126a08
15:44:02 <gammafunk> My guess is that delayed turns are not being counted properly, but I'm not sure how it's also inflating the turncount for actual qw
15:44:52 <gammafunk> the call to _update_place_info() happens at the end of _input() and delayed things and some kinds of handling of repeated commands return earlier in the loop
15:48:46 <elliptic> qw does weird stuff with macro delay so who knows
15:50:31 <gammafunk> Looks like I need to figure out that loop a little better before we can accurately track turns
15:50:38 <gammafunk> The xp and monsters counts should be correct, at least
15:50:51 <gammafunk> those simply happen when the player is awarded xp
15:51:45 <elliptic> can turn-tracking just happen when turncount is increased?
15:51:49 <gammafunk> |amethyst: since you asked about turns, see my URL http://dpaste.com/0GZ3D8F with the discrepancy between actual turns spent and the much smaller number the turn tracking in _update_place_info is reporting
15:51:57 <elliptic> also possibly aut-tracking would be better anyway
15:52:17 <gammafunk> yeah, I was thinking about that
15:52:48 <elliptic> though I guess spawns use regular turncount (that was what started this)
15:53:12 <gammafunk> right
15:53:32 -!- amalloy is now known as amalloy_
15:53:40 <elliptic> but it looks like you.num_turns only gets incremented in one place in main.cc
15:53:48 <elliptic> so could maybe just move stuff there?
15:54:36 <elliptic> (in world_reacts())
15:56:14 <gammafunk> hrm, let me try that
15:58:02 <|amethyst> gammafunk:  hm, odd... if you turn on turns_by_place, I assume it matches the real turncount, not the one shown there
15:58:21 <gammafunk> |amethyst: it doesn't actually
15:58:22 <elliptic> |amethyst: it didn't look right in the qw game earlier
15:58:29 <|amethyst> odd
15:58:45 <gammafunk> that little loop in _input() is a bit confusing
15:58:47 <gammafunk> there are a bunch of early returns
15:58:54 <gammafunk> with calls to _world_reacts()
15:59:01 <elliptic> I don't see any reason to put this in _input() anyway
15:59:15 <gammafunk> yeah, my only concern is some of the stuff in _update_place_info()
15:59:18 <gammafunk> is checking previous commands
15:59:36 <gammafunk> and whether the player is delayed
15:59:41 <elliptic> oh, because of the weird ignoring interlevel travel thing
16:00:03 <elliptic> IMO that isn't very useful anyway
16:08:41 <Cheibriados> 03MarvinPA02 07* 0.21-a0-205-ge9f4a53: Improve some berserk checks 10(59 seconds ago, 5 files, 16+ 20-) 13https://github.com/crawl/crawl/commit/e9f4a5387f11
16:09:04 <Jorgrell> Unstable branch on crawl.jorgrun.rocks updated to: 0.21-a0-204-ge2129ae (34)
16:09:12 <MarvinPA> guess who just spent a good 10 minutes wondering why on earth berserk was marked as useless when it shouldn't be, before realising maybe there was a missing "!"
17:09:07 <Jorgrell> Unstable branch on crawl.jorgrun.rocks updated to: 0.21-a0-205-ge9f4a53 (34)
17:09:26 <|amethyst> gammafunk:  since you know the item stats stuff better than I... do you have an estimate for what percentage of food fruit makes up (by number of items or by nutrition)?
17:09:47 <|amethyst> the future of Crawl depends on it
17:09:48 <gammafunk> uh, I think that's probably covered well by objstat
17:10:02 <gammafunk> good, finally the Fedhas removal crawl needs to be great again
17:10:04 <gammafunk> ??objstat[2
17:10:04 <Sequell> objstat[2/2]: Spreadsheets for all releases: https://drive.google.com/folderview?id=0B7VXhHzhWWb7S282VWhLVWRXbG8&usp=sharing ; See the README for details: https://docs.google.com/document/d/1D5mFqVi8ghz_nzvVmDUc3unx8VanVBWfgvZ8xCHaiJo/edit?usp=sharing
17:10:17 <gammafunk> Let me see
17:10:41 <|amethyst> gammafunk:  yeah, but I am in the middle of changing stuff so can't run objstat for a little bit
17:10:44 <|amethyst> :)
17:10:48 <gammafunk> |amethyst: the Comestibiles tabls should have what you need right there
17:10:51 <gammafunk> can you look at this sheet?
17:10:56 <gammafunk> https://docs.google.com/spreadsheets/d/1anrJhCbH2b71_Wtbv8wpdZdVw0yGaqPsKLzupy29Wd4/edit#gid=805322329
17:11:00 <gammafunk> s/sheet/workbook
17:11:04 <gammafunk> \/
17:11:18 <gammafunk> /
17:11:18 <|amethyst> gammafunk:  thanks, that's what I need
17:11:18 <gammafunk> ah, that's how you do that
17:11:21 <Floodkiller> any opinions on this? https://github.com/Floodkiller/crawl/commit/f57a1291af33e599cbba46df87f18edaabf9d426
17:11:29 <gammafunk> this is assuming you got shoals+spider
17:11:49 <gammafunk> which I don't think Lair branches would affect permafood generation too much
17:11:52 <|amethyst> gammafunk:  do you think anyone will be penalised too severely by the removal of the guaranteed food vaults?
17:12:08 <gammafunk> I can't imagine they would be
17:12:13 <gammafunk> You specifically mean the ones in Dungeon?
17:12:14 <|amethyst> cool
17:12:39 <gammafunk> I guess there are vaults in Lair that place a bunch of fruit, but I suppose it's hard to imagine that mattering to anyone save for Fedhas worshipers
17:13:17 <gammafunk> Food can matter for spellcasters who tend to undertrain spellcasting (and likely have lower int), and to gozag worshipers maybe potentially, but I think most food issues in crawl happen from spellcasting
17:13:26 <|amethyst> yeah, the uniq_first_food and chance_second_food vaults
17:13:56 <|amethyst> I was going to remove the former (though I could be persuaded to keep it) and fold the latter into random vault generation rather than having a CHANCE:
17:14:13 <gammafunk> Right, outside of challenge runs I don't think those matter; I've felt sorta a bit low on permafood in very early levels a few times, but it never really has gotten out of hand for me in any run
17:14:19 <|amethyst> hm
17:14:29 <|amethyst> maybe keep first_food then and remove second_food
17:14:48 <|amethyst> I am merging food down to either two or three subtypes, depending on whether I can reasonably make Fedhas use rations rather than fruit
17:15:03 <gammafunk> When I was arguing with hellmonk about food issues on spellcasters, that was just pointing out more longterm problems that can happen when you go with lowish spellcasting and extensive spell usage for a while; you can start to run low on permafood
17:15:16 <gammafunk> cool
17:16:02 <gammafunk> hrm, not sure if fedhas would work very well with rations as a cost
17:17:03 <|amethyst> gammafunk:  with the right number of rations as the cost it could be the same on average
17:17:23 <|amethyst> gammafunk:  ignoring carnivores and the spikiness of fruit vaults
17:17:43 <|amethyst> gammafunk:  hence my asking about food objstat
17:18:11 <gammafunk> hrm, but how would fedhas food costs work then? You mean that when e.g. creating a plant or evolving a plant, it costs an entire ration?
17:20:33 <|amethyst> average of 56 fruit per game, 35 meat, 36 bread, 35 RJ
17:21:00 <|amethyst> which will become something like 56 fruit, 85 rations with the changes I have already made
17:21:04 <|amethyst> (locally)
17:21:49 <gammafunk> oh, you're keeping fruit
17:21:54 <|amethyst> so far
17:21:55 <gammafunk> I was assuming fruit was going
17:22:08 <|amethyst> so if each fruit is replaced with (on average) 0.2 rations, and Fedhas requires 2 fruit per activation, it should be the same number of fruit per game
17:22:23 <|amethyst> but it costs a bit more nutrition unless you're a carnivore
17:22:36 <|amethyst> s/the same/close enough to the same/
17:23:15 <|amethyst> gammafunk:  commits so far, before handling fruits:
17:23:20 <|amethyst> Make food restriction mutations single-
17:23:22 <|amethyst> level.
17:23:29 <|amethyst> Simplify food_def.
17:23:33 <|amethyst> Merge bread and meat rations.
17:23:36 <|amethyst> Remove royal jellies.
17:25:55 <gammafunk> I guess I'm confused by: "if each fruit is replaced with (on average) 0.2 rations" if fruit is staying, why is it being replaced with rations?
17:26:12 <|amethyst> I haven't removed fruit yet, but I would like to
17:26:31 <|amethyst> so I'm trying to figure out numbers etc for doing so
17:26:57 <gammafunk> ok so you're arguing just that we can not disrupt fedhas if we switch his powers to cost rations
17:26:57 <|amethyst> yes
17:26:58 <gammafunk> if we replace fruit by the right amount of rations
17:27:02 <|amethyst> with the appropriate scaling
17:27:05 <|amethyst> yeah
17:27:12 <gammafunk> makes sense
17:30:44 <Ultraviolent4> are you all aware that mantis registrations are broken?
17:31:23 <|amethyst> Ultraviolent4: yes  https://crawl.develz.org/mantis/view.php?id=11182  suggestions or captcha improvements welcome
17:31:39 <|amethyst> Ultraviolent4:  until then, keep reloading until you don't get any of the inverse-video fs
17:32:47 <Ultraviolent4> thanks
17:32:56 <gammafunk> well, time to see if old qw HuBe can make it to Zot again
17:33:07 <gammafunk> oh I should test with a single-level morgue first
17:33:14 <|amethyst> btw, that captcha thing goes for everyone
17:33:29 <gammafunk> I moved _update_place_info to _world_reacts() like elliptic suggested, looking at the code it seems fine
17:33:41 <gammafunk> Yeah I saw Nap.kin's message about that, not sure what we can do
17:34:26 <Ultraviolent4> yea, im trying to solve a capthcha myself now lol
17:34:43 <Ultraviolent4> what am i looking for when i refresh it? all of them are the same type as far as i can tell
17:34:52 <|amethyst> gammafunk:  could move to github issues, but
17:35:06 <|amethyst> gammafunk:  this also affects tavern, since mantis and tavern use the same logins
17:35:07 <gammafunk> I have not seen any of those awful korean bot posts on tavern in a while, so at least there's some upside
17:35:35 <gammafunk> right, we need that to work well for taven, don't we? Or does tavern registration work through a different UI
17:36:00 <gammafunk> oh, I see, it's the same thing
17:36:37 <gammafunk> Nap.kin has made a captcha so difficult, not even google Deep Mind can solve it
17:37:38 <Ultraviolent4> yea, i was told by someone who was trying to sign up for tavern
17:38:00 <johnstein> what are the major issues with using github issues, except for the whole lack of knobs and dials thing?
17:38:36 <|amethyst> johnstein:  old bugs wouldn't be ported so we'd have two places to search
17:38:45 <|amethyst> unless we found a way to automate moving those
17:39:09 <gammafunk> I think there are some significant UI problems too; we discussed this a couple times yet I don't remember all the details, but isn't the way they handle templates not very good?
17:45:41 <johnstein> I think there's a way to automate github issue creation but I've never done it
17:45:48 <johnstein> though
17:45:57 <johnstein> scraping mantis might not be fun
17:47:45 <gammafunk> yeah, scraping wouldn't do at all, would have to be an import
18:14:15 <Lantell> Unstable branch on underhound.eu updated to: 0.21-a0-205-ge9f4a53 (34)
18:46:45 <Cheibriados> 03|amethyst02 07* 0.21-a0-206-g7d95abb: Checkwhite. 10(4 hours ago, 1 file, 8+ 8-) 13https://github.com/crawl/crawl/commit/7d95abb50f06
18:46:45 <Cheibriados> 03|amethyst02 07* 0.21-a0-207-g4a9efaf: Remove unused descriptions. 10(6 minutes ago, 16 files, 0+ 64-) 13https://github.com/crawl/crawl/commit/4a9efaf00671
18:59:56 <Cheibriados> New branch created: food-unification (5 commits) 13https://github.com/crawl/crawl/tree/food-unification
18:59:57 <Cheibriados> 03|amethyst02 07[food-unification] * 0.21-a0-208-gd1b9bdd: Make food restriction mutations single-level. 10(4 hours ago, 14 files, 75+ 79-) 13https://github.com/crawl/crawl/commit/d1b9bddb05e6
18:59:57 <Cheibriados> 03|amethyst02 07[food-unification] * 0.21-a0-209-gc400fd8: Simplify food_def. 10(3 hours ago, 2 files, 15+ 22-) 13https://github.com/crawl/crawl/commit/c400fd824c97
18:59:57 <Cheibriados> 03|amethyst02 07[food-unification] * 0.21-a0-210-g59b34a8: Merge bread and meat rations. 10(3 hours ago, 51 files, 129+ 297-) 13https://github.com/crawl/crawl/commit/59b34a872405
18:59:57 <Cheibriados> 03|amethyst02 07[food-unification] * 0.21-a0-211-g1e659e2: Remove royal jellies. 10(2 hours ago, 43 files, 89+ 144-) 13https://github.com/crawl/crawl/commit/1e659e220f7f
18:59:57 <Cheibriados> 03|amethyst02 07[food-unification] * 0.21-a0-212-g9efcdf6: Merge fruit into rations. 10(40 minutes ago, 38 files, 122+ 180-) 13https://github.com/crawl/crawl/commit/9efcdf613cfb
19:09:10 <Jorgrell> Unstable branch on crawl.jorgrun.rocks updated to: 0.21-a0-207-g4a9efaf (34)
19:19:42 <gammafunk> ho boy
19:19:48 <gammafunk> qw got the obsidian axe in lair
19:20:16 <gammafunk> aw, rip
19:20:26 <gammafunk> can't ever get your hopes up when qw plays HuBe I guess
19:43:00 <travis-ci> The build has errored. (food-unification - 9efcdf6 #8637 : Neil Moore): https://travis-ci.org/crawl/crawl/builds/266388397
20:55:15 <gammafunk> elliptic: Here's a HuBe that got to Depths:4 with the fixed turn tracking. http://dpaste.com/0VEXVKA
20:55:28 <gammafunk> I'll probably clean this up and push it tomorrow with the tabled enabled in trunk
20:55:51 <gammafunk> Not sure if making it mandatory in the dump is necessary, probably just enable it in the default config
21:05:16 <johnstein> |amethyst: will we want an experimental for food unification?
21:05:16 <elliptic> enabling it by default for a little while sounds good
21:50:45 <johnstein> how does this work?     for (monster_iterator mi; mi; ++mi)  I'm not really familiar with this kind of approach. looks like it's a way to pass a variable that will always iterate through a list of monsters anywhere it's used later
21:51:07 <johnstein> i.e. shorthand foreach monster in whatever list it's using
21:51:32 <geekosaur> !source monster_iterator
21:51:33 <Sequell> 1/2. https://github.com/crawl/crawl/blob/master/crawl-ref/source/act-iter.cc#L164
21:52:00 <johnstein> yea, I was looking at that. still trying to wrap my head around it fully
21:53:04 <geekosaur> just goes through the level's monster list
21:53:09 <geekosaur> skipping dead monsters
21:53:14 <johnstein> is it per level? or the entire game?
21:53:45 <johnstein> or maybe only the current level *IS* the entire game, at that moment
21:53:56 <geekosaur> you can only get other levels by loading them
21:53:59 <johnstein> (based on what |amethyst was explaining to me earlier)
21:54:41 <geekosaur> so yes, the current level is all that is known at that time; to work with monsters on another level, you load it (which saves the current level)
21:54:51 <geekosaur> then you reload the current level when done
21:55:11 <geekosaur> there's a flag for level loading that says it's only being visited for e.g. monster movement while offlevel, so don't force redraws etc.
21:57:05 <geekosaur> obviously this means you can't switch levels like that while iterators are current, because they'll all be invalidated
22:12:53 <advil> hm maybe we should use console dcss symbol meanings for the captcha
22:12:59 <advil> @1FuQ = player demon frog ugly thing tengu
22:19:31 <gammafunk> well it'd certainly be easy to game if anyone wanted to
22:19:48 <gammafunk> also vast majority of players don't know glyphs
22:19:59 <johnstein> rip tiles bug reporters
22:20:21 <gammafunk> we could show tiles instead, and screw over the ~8% of players who are console
22:20:46 <gammafunk> well we get pretty few new console players
22:21:22 <johnstein> just have a reload button for console and one for tiles
22:21:57 <CanOfWorms> could have the captcha be a single letter and ask the player to spell the god's name
22:22:04 <|amethyst> remember that the goal is to tell whether someone is human, not whether they are sufficiently knowledgeable about Crawl
22:22:06 <CanOfWorms> the only letter the captcha gives is K though
22:23:06 <|amethyst> we probably don't want to exclude "I just downloaded this program but it crashes when I try to run it" reports
22:23:35 <johnstein> show one of those new captchas with the multiple panes. click the panes with monsters susceptible to fire
22:23:39 <CanOfWorms> what about "what is the full word for C:S in DC:SS"
22:23:49 <johnstein> click all panes with monsters that mutate you
22:29:15 <gammafunk> |amethyst: so we could have an alternate that quizzes from the Jargon file if they can't pass the DCSS quiz?
22:29:45 <gammafunk> (Don't let game devs design your captcha)
22:30:41 <gammafunk> Beat this Contra NES stage in order to register
23:25:38 <Cheibriados> 03|amethyst02 07[food-unification] * 0.21-a0-213-gcd6f74d: Fix swapped herb/carn food values. 10(32 seconds ago, 1 file, 2+ 2-) 13https://github.com/crawl/crawl/commit/cd6f74db1fba
23:38:52 -!- amalloy_ is now known as amalloy
23:44:11 <amalloy> imo the captcha should be something like http://photobucket.com/gallery/http://s1218.photobucket.com/user/stinksandbangs/media/Port%20Rowan%20in%20S/Resources/Calvin-TreeFort_zpsffcc6a34.jpg.html but about the dev team
23:44:49 <amalloy> man that link sucks. let me find a direct link
23:45:26 <amalloy> https://s-media-cache-ak0.pinimg.com/originals/1b/25/6d/1b256d85301d0127f747c0e6484f3d95.jpg
23:58:51 <minmay> please draw your phone number on the minimap using this formicid