00:20:34 Unstable branch on cbro.berotato.org updated to: 0.29-a0-270-g1586fc07ea (34) 00:54:12 Monster database of master branch on crawl.develz.org updated to: 0.29-a0-270-g1586fc07ea 01:03:20 New branch created: pull/2477 (1 commit) 13https://github.com/crawl/crawl/pull/2477 01:03:20 03RojjaCebolla02 07https://github.com/crawl/crawl/pull/2477 * 0.29-a0-235-g8ecfb24: Simplify job starting stats 10(33 minutes ago, 1 file, 33+ 24-) 13https://github.com/crawl/crawl/commit/8ecfb24ed0f5 01:40:34 Fork (bcrawl) on crawl.kelbi.org updated to: 0.23-a0-4567-gb362bb5a12 02:05:17 -!- indigaz9 is now known as indigaz 04:23:04 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-4567-gb362bb5a12 05:07:08 Unstable branch on crawl.akrasiac.org updated to: 0.29-a0-270-g1586fc0 (34) 05:54:55 By the way, how does slow movement speed affect opportunity attacks (if at all)? 05:58:59 <12ebering> The same as fast monsters 06:00:31 ah okay, cool 08:02:05 -!- aoei is now known as stella 10:59:52 <12ebering> Re, Yermak's feedback. Yermak is here so we can get his full thoughts and not just one screenshot 11:00:07 <12ebering> hate can be: this is not fun or good gameplay; or that this is the wrong difficulty 11:00:08 <12ebering> @PleasingFungus 11:01:19 <10PleasingFungus> good point! 🙂 11:01:27 <10PleasingFungus> would be good to hear more than a thirdhand screenshot 11:02:28 <10PleasingFungus> reposting my comment from the other discord for context 11:02:28 <10PleasingFungus> i do think it'd be good to put more thought into this. If we want to make some parts of the game easier post-0.28 and/or to keep difficulty more or less similar pre and post AOOPs, what do we want to relax on? it's certainly possible to win the game without any player buffs or dungeon nerfs or w/e regardless, but that's not quite the same as the difficulty being "ideal" 1 11:02:33 <10PleasingFungus> ...1... 11:02:34 <10PleasingFungus> anyway 11:05:02 <10PleasingFungus> it's sort of weird because the extra difficulty in 0.27 and 0.28 seems quite diffuse - nothing in particular (branch or monster) jumped in kill rate 11:05:02 <10PleasingFungus> we already made some changes - nerfs to spider and D monster gen, for example 11:05:02 <10PleasingFungus> and we could certainly let those settle before doing more 11:05:02 <10PleasingFungus> but it'd be interesting to think about: if we wanted to increase winrate, what would be the most fun way to do it? 11:05:02 <10PleasingFungus> without making the "average fight" boring 11:05:02 <10PleasingFungus> does it involve giving players more consumables and actives, or better ones? does it involve changing monsters in some way? 11:05:02 <10PleasingFungus> or level gen? 11:05:02 <10PleasingFungus> i don't have particular answers, just think it's worth ruminating on. like an herbivore 11:05:02 <09gammafunk> yes 11:05:02 <09gammafunk> and there's a very real chance that the problem has to do with mimics 11:05:38 <09gammafunk> I guess I need more aoo games under my belt before suggesting anything 11:05:55 <09gammafunk> and yeah we didn't really nail down just why players died more in 0.28 than we'd have thought, either 11:06:10 <10PleasingFungus> oh, you're right. the problem is that there should be more mimics 11:06:17 <10PleasingFungus> then difficulty would drop 11:06:22 ebering, what are you talking about? 11:07:01 <10PleasingFungus> someone posted a screenshot of you saying something like "i don't like the new system" 11:07:04 <10PleasingFungus> i'm trying to find it 11:07:07 <10PleasingFungus> i have no idea what the context was 11:07:20 Oh, I was talking about how I don't like closing attacks 11:07:50 <10PleasingFungus> ah ha 11:07:51 <10PleasingFungus> https://cdn.discordapp.com/attachments/747522859361894521/950454747339587664/unknown.png 11:08:00 <10PleasingFungus> yeah, so that's what we wanted to ask about, i think! 11:08:21 <10PleasingFungus> as ebering said above 11:08:53 sec 11:11:53 !lg * t0.27|t0.28 !boring / mage s=cv 11:11:53 36444/136408 games for * (((t0.27 || t0.28)) !boring): 19283/67408x 0.28 [28.61%], 15431/63041x 0.27 [24.48%], 1240/4166x 0.27-a [29.76%], 490/1793x 0.28-a [27.33%] 11:11:53 that's the only bit of signal I've found re 0.27 vs 0.28 11:11:53 <10PleasingFungus> !lg * t0.27|t0.28 !boring / mage s=cv o=cv 11:11:53 otherwise it seems likely that the effect is an across the board one 11:11:53 <04Cerebot> 36444/136408 games for * (((t0.27 || t0.28)) !boring): 19283/67408x 0.28 [28.61%], 490/1793x 0.28-a [27.33%], 15431/63041x 0.27 [24.48%], 1240/4166x 0.27-a [29.76%] 11:12:02 Can you potentially blame shield changes? 11:12:04 <09gammafunk> recommend the kw byverions 11:12:07 <09gammafunk> for those types of queries 11:12:09 <10PleasingFungus> i'm having a hard time parsing this.... 11:12:15 <09gammafunk> er byversion 11:12:20 <10PleasingFungus> !kw byversion 11:12:21 <04Cerebot> Keyword: byversion => s=regexp_replace(cv, "-a", "") 11:12:21 <09gammafunk> instead of s=cv 11:12:29 <10PleasingFungus> s=byversion(cv)? 11:12:33 <09gammafunk> nope 11:12:34 <09gammafunk> just the kw 11:12:34 <10PleasingFungus> oh i see 11:12:45 <10PleasingFungus> !lg * t0.27|t0.28 !boring / mage s=byversion 11:12:50 <04Cerebot> Unknown field: byversion 11:12:53 <10PleasingFungus> oops 11:12:54 just byversion 11:12:57 <10PleasingFungus> !lg * t0.27|t0.28 !boring / mage byversion 11:13:01 <10PleasingFungus> yeah i knew that, just messed up when typing 11:13:04 <10PleasingFungus> doh 11:13:05 <04Cerebot> 36444/136408 games for * (((t0.27 || t0.28)) !boring): 19773/69201x 0.28 [28.57%], 16671/67207x 0.27 [24.81%] 11:13:28 <09gammafunk> finally, casters beat melee? 11:13:29 <10PleasingFungus> ok, so we've learned that people played more games as mages in 0.28? 11:13:36 <10PleasingFungus> this shows nothing about winrate, just play rate 11:13:47 <10PleasingFungus> it is interesting 11:13:51 well, I think you aren't thinking it out -- more people playing mages => more mage deaths => lower winrate 11:13:57 <10PleasingFungus> oh, heh 11:14:09 I didn't actually check the baserate tbh 11:14:11 <10PleasingFungus> assumption is that mages win less in general? they're no berserkers, for sure 11:14:19 <10PleasingFungus> !lg * !boring / mage byversion -graph 11:14:25 !lg * t0.27|t0.28 !boring mage byversion 11:14:26 <04Cerebot> 3397237/12632699 games for * (!boring): https://betelgeuse.shalott.org/graphs/a476fb022a51f812046d0ca38a34d3db15a6c2d9.html 11:14:30 36444 games for * (((t0.27 || t0.28)) !boring mage byversion): 19773x 0.28, 16671x 0.27 11:14:38 so definitely a lot more mage games 11:14:40 <10PleasingFungus> oh no this graph is bad... 11:14:48 <09gammafunk> can add o=. 11:14:49 for whatever reason (maybe dj?) 11:14:52 <10PleasingFungus> !lg * !boring / mage byversion o=. -graph 11:14:59 <04Cerebot> 3397241/12632710 games for * (!boring): https://betelgeuse.shalott.org/graphs/63f6d8a9bc52f31870b9ca886b8abcd5e7d5d593.html 11:15:14 <09gammafunk> I guess you'd need to exclude 0.1 weirdness, but that's better 11:15:26 <09gammafunk> oh that's like 11:15:29 <09gammafunk> cv= 11:15:31 <10PleasingFungus> nothing really jumps out here... 11:15:32 <09gammafunk> wonder what those are 11:15:44 <10PleasingFungus> !lg * cv= 11:15:48 <04Cerebot> 5. lunkers the Geomancer (L11 VpEE of Ashenzari), blasted by Fannar (bolt of cold) on D:10 on 2021-01-11 18:15:52, with 9528 points after 13139 turns and 0:23:11. 11:15:50 <10PleasingFungus> hm 11:15:54 <09gammafunk> weird 11:15:56 <10PleasingFungus> mystery games 11:15:57 <09gammafunk> that's recent 11:15:58 !lg * t0.27|t0.28 !boring / !mage byversion 11:16:01 <10PleasingFungus> !lg * cv= s=vlong,src 11:16:07 99964/136408 games for * (((t0.27 || t0.28)) !boring): 50536/67207x 0.27 [75.19%], 49428/69201x 0.28 [71.43%] 11:16:08 <04Cerebot> 5 games for * (cv=): 0.19-a0-856-g54f7fbd (cwz), 0.20.1-1-g8b1a95d (cao), 0.20-a0-739-g6bb112f (cbro), 0.22-a0-425-g89d60a1 (cao), 0.27-a0-54-g64bdb92e38 (cbr2) 11:16:12 <09gammafunk> !lg * cv= x=file 11:16:14 <04Cerebot> 5. [file=cbr2/meta/git/logfile] lunkers the Geomancer (L11 VpEE of Ashenzari), blasted by Fannar (bolt of cold) on D:10 on 2021-01-11 18:15:52, with 9528 points after 13139 turns and 0:23:11. 11:16:15 <10PleasingFungus> weird 11:16:22 for !mage backgrounds the trend seems reversed in fact 11:16:24 <09gammafunk> hrm that seems bugged 11:16:42 <10PleasingFungus> advil: isn't that tautological? 11:16:54 <09gammafunk> yeah 11:18:44 <09gammafunk> necessity, right? 11:18:44 no? it could be higher death rate for both in 0.28 11:18:44 <10PleasingFungus> you haven't seen anything about death rate 11:18:44 <09gammafunk> that's not death rate advil 11:18:44 <09gammafunk> that's play rate 11:18:44 <10PleasingFungus> there is no info about death rate in either query 11:18:44 haha right, sry 11:18:44 <10PleasingFungus> !lg * t0.27|t0.28 !boring mage byversion / won 11:18:44 <10PleasingFungus> !lg * t0.27|t0.28 !boring !mage byversion / won 11:18:44 <04Cerebot> 853/36444 games for * (((t0.27 || t0.28)) !boring mage byversion): 448/16671x 0.27 [2.69%], 405/19773x 0.28 [2.05%] 11:18:44 <04Cerebot> 1879/99964 games for * (((t0.27 || t0.28)) !boring !mage byversion): 1124/50536x 0.27 [2.22%], 755/49428x 0.28 [1.53%] 11:18:44 <10PleasingFungus> look at em go 11:18:44 but yeah, it does hold up I think 11:18:44 <09gammafunk> so it is lower for non-mages too 11:18:44 <10PleasingFungus> 📉 11:18:44 <10PleasingFungus> sorry, what holds up? 11:18:44 !lg * t0.27|t0.28 !boring !won / mage byversion 11:18:44 !lg * t0.27|t0.28 !boring !won / !mage byversion 11:18:45 35591/133676 games for * (((t0.27 || t0.28)) !boring !won): 19368/68041x 0.28 [28.47%], 16223/65635x 0.27 [24.72%] 11:18:54 98085/133676 games for * (((t0.27 || t0.28)) !boring !won): 49412/65635x 0.27 [75.28%], 48673/68041x 0.28 [71.53%] 11:18:57 <10PleasingFungus> advil, that still doesn't show death rate 11:19:33 <10PleasingFungus> in fact, the death rate queries i just did suggest mages are won more than non-mages, so we'd expect that, if people hadn't shifted toward mages, the winrate drop might have been even steeper 11:19:37 <10PleasingFungus> (maybe! depending on your assumptions) 11:19:43 er, it doesn't? 11:19:48 <10PleasingFungus> it does not 11:19:57 <10PleasingFungus> the queries you just did just showed, out of all deaths, what percentage were from mages 11:20:07 <10PleasingFungus> it's not asking, out of all mage games, what percentage are deaths 11:20:10 <09gammafunk> yeah, the ratio syntax is a bit weird 11:20:17 <09gammafunk> the numerator is the term after the / 11:20:24 isn't the denominator all games for those queries? 11:20:29 <10PleasingFungus> it is not 11:20:29 <09gammafunk> denominator is what's before 11:20:40 <10PleasingFungus> the denominator is t0.27|t0.28 !boring !won 11:20:40 which does not include `mage` or `!mage` 11:20:49 <09gammafunk> nope, the way it works is terms before / apply to numerator and denominator 11:20:59 <09gammafunk> and terms after / apply only to numerator 11:21:18 <10PleasingFungus> that is, there were 68041 unwon unboring games in t0.28, of which 19368 (28.47%) were mages 11:21:31 I intended to show death rate, not winrate 11:21:45 <09gammafunk> it's somewhat confusing syntax, I guess, but it does sort of make sense for most type of ratio queries one wants to do, in that it's concise 11:21:52 tbc 11:21:59 <10PleasingFungus> i don't know what you're trying to express, then. that's not very meaningful data. 11:22:00 <09gammafunk> otherwise you'd have to sort of verbosely fine numerator and denominator 11:22:16 <10PleasingFungus> it's approximately the same as play rate (which you'd expect, given a 2% overall winrate) 11:22:16 <09gammafunk> pf did death rates above 11:22:28 <10PleasingFungus> i guess it depends how you define "death rate" 11:22:56 <09gammafunk> oh, yeah, I don't know what "death rate" actually means, right he did winrate above 11:22:58 hm, is your point that it is only out of !won games? 11:23:03 <10PleasingFungus> let me rephrase 11:23:12 <10PleasingFungus> saying "out of X lost t games, Y of them were mages" doesn't tell us very much 11:23:18 <10PleasingFungus> it tells us that people were playing about that many mages 11:23:24 <10PleasingFungus> but it doesn't tell us anything about how hard mages are 11:23:33 <10PleasingFungus> or how much that difficulty changed over time 11:23:36 <10PleasingFungus> does that make sense? 11:25:37 <09gammafunk> we know that mages were played more systematically in 0.28, compared to 0.27 11:25:37 <09gammafunk> or at least I think we established that first, didn't we 11:25:37 <10PleasingFungus> in t0.28 11:25:37 <09gammafunk> right 11:25:37 <10PleasingFungus> don't see the same trend in overall data including non-t 11:25:37 <10PleasingFungus> but maybe we only care about t for now 11:25:37 <09gammafunk> but even when looking at non-mages only, winrate lower in t0.28 compared to t0.27 11:25:37 <09gammafunk> so it can't really be that it was people playing mages more 11:25:51 While I'm putting words together, I have a question, I read that slower enemies don't get closing attacks, by yesterday I was hit by orc zombies as normal speed character. 11:25:59 <10PleasingFungus> also, mage winrate is and was higher than non mage winrate 11:26:17 <10PleasingFungus> yermak: it's not currently true that slow enemies, though i'm considering making it true 11:26:28 oh, ok 11:26:41 So Spriggans and felids have harder time than I thought? 11:26:42 <10PleasingFungus> oops, forgot an important word in that sentence 11:26:44 <10PleasingFungus> but you get the idea 11:27:29 <09gammafunk> !lg * t0.28 byrolegroup o=% / won 11:27:31 <04Cerebot> 1160/75041 games for * (t0.28 byrolegroup): 356/17920x Mage [1.99%], 111/6781x Adventurer [1.64%], 175/10796x Zealot [1.62%], 113/8055x WarriorMage [1.40%], 405/31489x Warrior [1.29%] 11:27:55 <09gammafunk> !lg * t0.27 byrolegroup o=% / won 11:27:59 <04Cerebot> 1572/71401 games for * (t0.27 byrolegroup): 379/14886x Mage [2.55%], 165/7296x Adventurer [2.26%], 181/8247x WarriorMage [2.19%], 635/30496x Warrior [2.08%], 212/10476x Zealot [2.02%] 11:28:15 <09gammafunk> yeah nothing too surprising there I guess 11:28:55 well, it's not an even drop across the roles, I guess in a direction contra to what I was originally suggesting 11:29:07 hard to say if those are different though 11:30:33 <09gammafunk> yeah not the same drops, but the spread in drops doesn't seem extreme 11:32:27 <09gammafunk> from .45% to .8% or so 11:32:27 <09gammafunk> oh 11:32:27 <09gammafunk> that graph I did a while back of difficulty by xl might help 11:32:27 the player meme is that melee is impossible now (even before recent changes), which is sort of in line with the warrior difference being the biggest 11:32:27 <09gammafunk> unfortunately I'm a bit out of time, that takes a few queries to set up 11:33:11 When did the 'monsters ignore weapon delay' change come about? 11:33:57 <09gammafunk> https://cdn.discordapp.com/attachments/747522859361894521/950461319033810954/dcss_difficulty_0.27.png 11:34:04 <09gammafunk> this image, could make this again for 0.28 11:34:12 <12ebering> yes that would be helpful 11:34:42 <09gammafunk> if anyone wants to look up how I did that and graph, feel free, I can even share the ods I used to make it, but I do have to go now 11:34:50 <09gammafunk> otherwise I can make tomorrow morning 11:43:51 <12ebering> if its not too much trouble to post a quick summary I can take a crack 11:43:59 <12ebering> (recognizing that you might already be gone for now) 11:53:10 I'm not sure yet, if my dislike is mostly the reluctance to adopt new tactics system or not. 11:53:10 When it was all talk about closing attacks (CA), I didn't read much into it and was under impression you got rid of the kiting by giving players guaranteed opportunity to get the space between them and enemy by eating one attack. It turned out nothing like that (well, maybe not for ogres). 11:53:10 1. I think the amount of kiting might have increased in my games. It's just now kiting with a space between you and enemy. How it was earlier for casual (more or less) experienced players (I think): you fight your enemy where you met it and if someone else appears (I'll ballpark chances at 20%), you reposition into a choke point. Now you can't do it, and have to reposition for each not absolutely trivial fight. 11:53:10 2. When you hold tab in a choke point and sometimes stick out of your den, you get twofold punishment now - both when sticking out and when trying to get back. 11:53:10 3. The difference between kiting an enemy that's close to you and that's not is unnaturally sharp. 11:53:10 4. I haven't played Sp and Fe yet, but I imagine kiting with them is a lot more tedious now - you should always keep your enemy two tiles away. More so with range limited attacks, like Scorch. 11:53:17 <12ebering> ty for the detailed thoughts 11:57:22 <09gammafunk> no time for quick summary, but the method to calculate "difficulty" is simply take deaths by s=xl D_1,..,D_27 and games by s=xl G_1,...,G_27, then "difficulty" for XL N is D_N / sum(G_N,G_27) 11:58:03 I suggest the following: make every new enemy take their time to adjust to your step - you can make free dozen steps (they could be divided apart in time) back before it starts performing CA. It will solve 1 and 2. 11:58:04 <09gammafunk> and by those values D_N and G_N I just mean the counts 12:00:07 <09gammafunk> and add conditions to filter for non-extended etc 12:00:07 <12ebering> games is all games that ended at that xl or some milestone business? 12:00:07 did you use only goodplayers for the denominator, or all games? 12:00:07 Or another (less favoured by me) potion: you don't get CA while you're at full HP (while you don't limp). 12:00:07 option* 12:00:07 <09gammafunk> yeah just games for both counts, not milestones at all 12:00:07 <09gammafunk> the filter in that graph was players with 10+ wins and only tournament games, applied to both sets of values 12:00:07 <09gammafunk> could remove said goodplayer filter 12:00:07 <09gammafunk> maybe not relevant to what we're after 12:00:07 (asking because a lot of @goodplayer queries are timing out for me) 12:00:38 <12ebering> what sort of extended filter did you use 12:00:44 <09gammafunk> "difficulty at xl N" is "of the games that made it to XL N or higher, which proportion died at xl N" 12:01:01 <12ebering> (I guess I could make the line for several versions with a consistent extended filter) 12:01:54 <09gammafunk> > For 0.26 and 0.27 tournaments, the proportion of games losing at each XL out of all games that made it to that XL or further. This graph only includes non-quit games that got 3 runes or less and played by players with 10 or more wins before the respective tournament started. Games that ended in the Hells, Pan, Tomb, or a Zig are also excluded. 12:02:32 <09gammafunk> so looks like I excluded any game that ended in extended-only branches and also included urune<=3 and !boring 12:03:20 <12ebering> cool 12:03:21 <12ebering> thanks! 12:35:39 (Late, but the way I think of ratio queries is that !lg * a / b is just a is the less specific query while b is the more specific one and is calculated as a && b) 12:41:16 I wonder how difficult it'd be to expose Sequell's DB to a public-ish Postgres/whatever server or just to a downloadable archive for offline access 12:47:37 03RojjaCebolla02 {GitHub} 07https://github.com/crawl/crawl/pull/2466 * 0.29-a0-249-g17c2a6c: dump oklob mon-pick data 10(5 minutes ago, 1 file, 0+ 1-) 13https://github.com/crawl/crawl/commit/17c2a6c4eecd 13:01:58 e'bering, i added the tiles to the PR so that you don't have to use the github UI to see them 13:03:44 i went with tentacles because of a theme with horrors / the abyss, but vines work just as well. and tentacles kinda imply constrict. constrict is not particularly necessary, i just wanted to up the threat. i'll dump the constrict? 13:03:52 simpler design better and whatnot 13:22:38 <12ebering> ya the constrict is not doin much for them 13:23:12 i do think "don't use harpoon shot when you can melee" would be really good logic for that spell in general. swamp worms are better than they used to be but still pretty laughable 13:23:16 do any other spells have that logic already? 13:24:01 i guess centaurs work that way, self-gentleing if you get close enough for hugs 13:25:36 zanier fix: if something harpoon shots a player in melee, the player gets yanked to some other spot adj to the monster 13:29:34 <12ebering> There's a data structure (somewhere in mon-cast.cc) called mon_spell_logic and spell to logic used by monster ai to decide when to use a spell and on whom 13:29:34 <12ebering> it's newer, the lava layer below just targets a simple beam 13:29:34 <12ebering> harpoon shot currently uses the layer below, but a logic to not use it when we can melee the target would be reasonable 13:29:37 hopefully the ai already considers reach attacks to be "melee"? 13:30:19 <12ebering> yes 13:30:30 <12ebering> I mean, that's not a well posed question when you talk about code paths &c. 13:30:36 <12ebering> but, "yes" 13:31:58 <10PleasingFungus> yermak, thank you for the great writeup! re taking time to adjust, seems unfortunate in that it's more to track (both for the game & the player) & feels like it might encourage strange behavior, depending on how you mean "divided apart by time." How do you tell the difference between someone who's pillar dancing with occasional breaks to 'reset' the opportunity attack system vs someone who's just repositioning...? Sounds complex, 13:31:58 to me 13:32:08 <10PleasingFungus> regardless, i will keep thinking about things 13:33:02 <06advil> The harpoon shot thing should be possible though sadly monster ai isn’t really “consider all the options, score them given the circumstances, and choose among the best” 13:33:51 <12ebering> I think it's a feature, otherwise we'd probably end up with a few accidental smart kobolds 13:34:15 <12ebering> (for those who don't know Smart Kobold is a roguelike where the AI is Very Smart) 13:34:26 By the charming Jeff Lait IIRC 13:36:30 "the orb of fire hands you a questionnaire asking how much rF you have got and how likely you would be to ragequit if it malmutated you six times in a row" 13:36:30 <06advil> Actually spell choice is the closest to what I said 13:37:50 <10PleasingFungus> monsters being really dumb is often really great 13:37:57 <10PleasingFungus> the player feeling smart is wonderful 13:38:03 <10PleasingFungus> the player's allies feeling dumb, on the other hand... less great 13:38:06 <10PleasingFungus> the duality of game 13:38:40 advil> I had some inclination after fiddling with fireflies 13:43:27 <06advil> yeah, I don't really mean in the sense of making monsters smart, but there's not even a point in the code where it has a set of actions like {swing, cast 1, cast 2} and it checks whether all are possible 13:44:01 <06advil> monsters being smart inevitably leads to 4.1 alpha 13:44:46 <06advil> the spell ai does "score given the circumstances" where that score is basically "is it possible or not" 13:45:06 Also the questionnaire would burn up so my idea is not feasible 13:45:39 ah yeah, like how shady drags will burf at someone with rn+++, but not rNinfinity 13:53:23 Prerequisite 'mon-util.h' is newer than target 'cloud.o'. Must remake cloud.o <-- has anyone seen ccache doing a bunch of this sort of thing, and throwing out the cache? more importantly, do you just put up with it or is there a way to trick it into doing a better job 13:53:32 it's rebuilding tons of files that i haven't touched 13:54:51 Can you show a bit of context for that output appearing? 13:57:08 ah, like the whole build output in a pastebin? 13:57:08 or just, the fact that i'm rebase-squashing? 13:57:08 the former 13:57:29 sec 13:57:37 a lot of it scrolled past the terminal buffer limit V_V 13:58:08 https://0x0.st/oPGH.txt 14:00:21 I ask because that just looks like "make -d" saying it, and that would be the expected behaviour. 14:01:01 Have the "ccache -s" numbers changed since? 14:03:26 03RojjaCebolla02 07https://github.com/crawl/crawl/pull/2466 * 0.29-a0-236-ge3bf91c: Starflower monster enum 10(3 days ago, 2 files, 24+ 0-) 13https://github.com/crawl/crawl/commit/e3bf91cbd81a 14:03:26 03RojjaCebolla02 07https://github.com/crawl/crawl/pull/2466 * 0.29-a0-237-g2cc2ff8: mon-pick-data for starflower 10(3 days ago, 3 files, 13+ 0-) 13https://github.com/crawl/crawl/commit/2cc2ff8ef733 14:03:26 03RojjaCebolla02 07https://github.com/crawl/crawl/pull/2466 * 0.29-a0-238-gef86a58: dat/description for starflower 10(3 days ago, 1 file, 6+ 0-) 13https://github.com/crawl/crawl/commit/ef86a58f579f 14:03:26 03RojjaCebolla02 07https://github.com/crawl/crawl/pull/2466 * 0.29-a0-239-g53f8aac: Try to implement af_reach_disto as a range 3 slap 10(3 days ago, 7 files, 23+ 0-) 13https://github.com/crawl/crawl/commit/53f8aac52cb9 14:03:26 03RojjaCebolla02 07https://github.com/crawl/crawl/pull/2466 * 0.29-a0-240-ga90d56b: A couple quick starflower tiles 10(3 days ago, 4 files, 2+ 1-) 13https://github.com/crawl/crawl/commit/a90d56ba0e64 14:03:26 03RojjaCebolla02 07https://github.com/crawl/crawl/pull/2466 * 0.29-a0-241-g07d42b5: buff damage, change attack order 10(2 days ago, 1 file, 3+ 3-) 13https://github.com/crawl/crawl/commit/07d42b51b774 14:03:26 03RojjaCebolla02 07https://github.com/crawl/crawl/pull/2466 * 0.29-a0-242-gb2a443c: Revert "mon-pick-data for starflower" 10(13 minutes ago, 3 files, 0+ 13-) 13https://github.com/crawl/crawl/commit/b2a443c11b5f 14:04:02 oh! 14:04:02 i did switch to adding --debug to my build command a bit ago 14:04:02 that *always* rebuilds? 14:04:02 <12ebering> yes, debug sets all sorts of flags throughout the codebase 14:04:02 <12ebering> rebuilds basically everything 14:04:02 <12ebering> (maybe completely everything I don't use ccache) 14:04:02 No, it's just rebuilding it because the file modification times are as it says - which is expected since the rebase will have touched mon-util.h but not whatever.o 14:04:24 <06advil> if mon-util.h actually changed and the signature of cloud.cc + its included header files has actually changed and is in the cache, it doesn't have much choice 14:04:32 ebering> ah, which --debug is this? If it's make --debug that doesn't change the output. 14:04:40 I mean, the resulting foo.o 14:04:49 <06advil> if you consistently build with DEBUG=y then that gets cached, but maybe that's not what you mean? 14:04:53 last week i was using this same command, but without the --debug, sometime this week i added --debug so that i could do the stepinto thing to get a better idea of what the game does: ccache make -j 3 TILES=1 --debug && ccache -s && gitcraw 14:04:58 <06advil> *and isn't in the cache 14:05:18 <12ebering> oh, --debug isn't crawl's debug target, oops 14:05:36 spicyCebolla: What I'm getting at is everything you've shown is consistent with "make thinks it has to rebuild it because mon-util.h is newer because of the rebase, but ccache notices it's the same compile and serves up the results very quickly" 14:05:37 note thatwhat you are debugging there isnot crawl, but make. which is pretty useless 14:05:45 ahaha, i see 14:05:53 <06advil> in case it wasn't clear, mon-util.h is included by cloud.cc in a somewhat indirect way 14:05:55 geekosaur: make seems to be behaving entirely usefully 14:05:57 that explains why some of the stepinto's yield only strange info 14:06:16 spicyCebolla: So that's why I ask if ccache -s has changed - more cache hits, ccache doing its job 14:07:23 grand, ok. i don't have any prior outputs from ccache so i don't know. but i'll change that --debug flag to DEBUG=y now! 14:07:48 <06advil> cloud.cc -> dungeon.h -> env.h -> map-cell.h -> mon-info.h -> mon-util.h 14:08:01 <06advil> (it would be right to be a bit aghast at that) 14:08:03 amazing! 14:08:19 i mean it's one step closer to the ultimate simplification right ^_^ 14:08:26 (everything includes everything) 14:10:51 just put it all into AppHdr.h :) 14:10:51 pragma to the moon! 14:15:09 Oh, what was the other idea that was adjacent to OA along the lines of "if you make a move that would give an OA, that monster takes it if it has enough energy, /but the attack isn't free/" 14:15:39 E.g., One melee hit (per thing you're moving way from) > almost guaranteed one tile gap 14:17:44 I assume that wasn't gone for since that seems a bit too intense of a player buff to me, but curious to see if I'm missing something there with how energy mechanics works and all that jazz 14:17:44 It's more-or-less old WJC pin for everyone 14:17:44 (or, say, weapon swapping until monster swings then backing up) 14:44:35 03RojjaCebolla02 07https://github.com/crawl/crawl/pull/2466 * 0.29-a0-243-gab76d31: Implement many of ebering's improvements 10(9 minutes ago, 7 files, 16+ 29-) 13https://github.com/crawl/crawl/commit/ab76d31ea6b0 14:53:42 PF, sorry, missed your reply. There'd be no such thing as 'reset'. Twelve attempts to move away from a monster, divided apart by any number of turns - and voila, that monster starts to perform closing attacks on you. 1/3 chance can be even dropped. Imo, it's fulfills the goal of stopping kiting while allowing some repositioning. 14:56:11 "The -foo- adjusts to your coward style of combat and now will perform closing attacks on you as you try to escape." 14:56:53 Hm, we had some kind of monster-lifetime count idea too 14:58:54 senses your cowardice and closes in for the kill 14:59:18 <12ebering> %git dynamic-monsters 14:59:19 <04Cerebot> Could not find commit dynamic-monsters (git returned 128) 14:59:25 <12ebering> rip 14:59:54 <12ebering> %git dpeg_dynamic_monsters 14:59:54 <04Cerebot> Could not find commit dpeg_dynamic_monsters (git returned 128) 14:59:58 <12ebering> excuse you 15:00:02 <12ebering> oh 15:00:10 <12ebering> %git dpegs_dynamic_monsters 15:00:10 <04Cerebot> gammafunk * 0.21-a0-124-gab9bf4c: Merge remote-tracking branch 'origin/master' into dpegs_dynamic_monsters (4 years, 8 months ago, 0 files, 0+ 0-) https://github.com/crawl/crawl/commit/ab9bf4c84b99 15:01:27 <12ebering> Yermak: so you're suggesting that for every move away from the monster after the 12th it gets a closing attack? 15:01:53 yeah 15:12:17 <10PleasingFungus> coward style of combat, ha 15:12:17 <10PleasingFungus> seems annoying for players to count - optimal play is to count even before you hit the threshold of 12, right? 15:14:28 <12ebering> also how do you tell the difference between Y - a yak and Y - a yak but next time you back up its gonna start poking you 15:15:03 "Annoy count" as a status indicator /j 15:15:38 status light "Coward" :� 15:15:42 <10PleasingFungus> oh yeah, extra fun in console for sure 15:15:49 yup 15:15:52 <10PleasingFungus> hard problem 15:16:08 Make it not hard 12, but x+1dy 15:16:31 Still involves counting for x though 15:16:38 Hypothetically Optimal Man should still count the - what perryprog said 15:16:42 Also do you really need to care about such a status if you play normally without kiting? 15:17:05 <10PleasingFungus> think could be good to gentle the beast (opp attacks) a lil in some way. will keep ruminating 15:17:30 03advil02 07* 0.29-a0-271-gd4eb659: fix: allow anon spectating 10(2 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/d4eb659eea71 15:19:03 <10PleasingFungus> do wonder about making it a lil easier to create distance again, without going full wjc pin or annoying slow random processes like random energy 15:19:03 X / y chance of opp attack, where X is number of turns you have provoked an opp attack, and y is 10 or 12 or 15 or the current day of the month 15:19:03 turns in a row 15:19:03 <10PleasingFungus> double odds when jupiter is ascending 15:19:03 absolutely 15:19:03 <10PleasingFungus> half chance on leap years 15:19:03 I mean, nethack does that already! 15:19:18 Mercury in retrograde? Double chance of op attacks 15:32:29 Unstable branch on crawl.kelbi.org updated to: 0.29-a0-271-gd4eb659eea (34) 15:35:37 <12ebering> @PleasingFungus I'd let the experiment run for a little bit longer, give people some adjustment time 15:35:59 <12ebering> (I'm very strongly in the theoretical infinite free resets isn't actually good camp, fwiw.) 15:41:12 <10PleasingFungus> sure 15:41:18 <10PleasingFungus> i’m just ruminating 15:41:30 <10PleasingFungus> not planning to make any sudden drastic changes 15:41:56 <10PleasingFungus> just minor changes for now, and maybe ideas later 15:59:37 Pleasing-Fungus is actually Pleasing-Cow 16:02:16 apparently ruminants have all sorts of gut fungi? 16:02:37 https://www.sciencedirect.com/topics/agricultural-and-biological-sciences/rumen-fungi 16:13:18 Looks like this would be the main phylum: https://en.wikipedia.org/wiki/Neocallimastigomycota 16:13:18 symbionts, so they would indeed by a pleasing fungus 16:17:38 maddening error i keep not finding a solution to: https://0x0.st/oPGv.txt 16:18:04 i made flavour_has_bigreach look pretty much exactly like flavour_has_reach :\ 16:20:48 Isn't has_attack_flavour a property of monsters not attacks? 16:21:47 hence attk.has_attack_flavour() will never work and it's nothing to do with it being called on AF_RIFT 16:23:51 Unstable branch on underhound.eu updated to: 0.29-a0-271-gd4eb659eea (34) 16:24:47 * Pinkbeast says spicyCebolla in case your IRC client pings you 16:26:55 ooo hi, i'm also packing for a 40 hour busride and 2 week work-cation 16:26:55 am hoping i finish this PR on the train 16:26:55 s/bus/train 16:26:55 not as good as PF's baby excuse! 16:27:54 if (has_attack_flavour(af_foo)) is all over monster.cc though 16:27:59 which is where i'm using it 16:28:23 right, but it's presumably being used in a method of monster, not attack 16:28:37 (`this` is not what you think it is) 16:30:21 okay. when i understand this i'll try something new. thanks for the tips! pushing for now so i don't lose track of where i'm at 16:31:40 spicyCebolla: But the only function defined is "bool monster::has_attack_flavour(int flavour) const" so we know this is a function for asking about monsters not attackds 16:33:09 03RojjaCebolla02 07https://github.com/crawl/crawl/pull/2466 * 0.29-a0-244-gbc24ced: attempted fix for the reach_three problem 10(26 minutes ago, 4 files, 16+ 8-) 13https://github.com/crawl/crawl/commit/bc24cedfee6d 16:33:45 ah, are monsters not supposed to have multiple reach ranges per attack? (if so, i'm so confused why this worked before) 16:40:43 spicyCebolla: orange demons have AF_REACH_STING on one attack 16:41:10 But that's tested for with attk.flavour 16:41:43 So no, that's not a rule - it's just that what you're using is a check to see if a monster has _any_ attack with a given property 16:41:56 -!- TAS-2012v is now known as TAS_2012v 16:42:29 (eg many of those has_attack_flavour()s are to work out if Recite thinks the monster is chaotic / unclean / etc) 16:43:34 ah, i did try "if (attk.flavour == AF_RIFT" in a couple ways also 16:47:22 That should in principle work provided it's somewhere we know what "attk" is 16:47:22 I'm a bit confused as to why there are pull requests for something that doesn't build 16:47:22 But you could replace flavour_has_bigreach(attk.flavour) with (attk_flavour == AF_RIFT) 16:47:47 confused> FTAOD I am not a vanilla developer and, as such, this is not a coded way of saying "stop" 16:49:49 <06advil> I think this PR is in the "responding to review" stage 16:51:26 right, it was building, then with the review, i tried to get rid of a function that was kinda silly (attack_has_disto), and now it's not working, even though the simplified code is incredibly similar 16:51:34 i'll diff the working and nonworking versions on the train 16:54:47 <06advil> fwiw I don't know if flavour_has_bigreach is necessary but if I were implementing it, I would probably do it as: return flavour == AF_RIFT 16:55:01 <06advil> with a ; ofc 16:55:24 <06advil> after that commit it looks like it should work to me 16:55:53 <06advil> 40 hours is a lot of busriding 16:57:57 s/bus/train 16:58:23 ... it's also a lot of trainriding but I infer possibly on America's hilariously slow trains 16:58:35