IRC logs for #openttd on OFTC at 2019-01-15
⏴ go to previous day
00:00:05 <Samu> i like the idea of caching path
00:00:21 <peter1138> That's what I asked you to attempt to do last year.
00:00:26 <Samu> keep working on it, it was also something I had in mind, but with limited code skills I didn't bother
00:00:57 <peter1138> I think it did actually sort of work, but kept crashing.
00:03:15 * LordAro is better at paint than andy
00:04:24 <peter1138> Just make ships behave like triremes in civilization. If they get lost, they actually get lost, and disappear...
00:05:00 <peter1138> o_O ... this patch adds a std::stack<Trackdir> path to struct Ship.
00:05:10 <peter1138> Well I can try it :)
00:05:41 <LordAro> well, depends on its usage :p
00:09:06 <Samu> er, i made a review, maybe i wanted to post single comment, my bad
00:09:10 <nielsm> as the name says, it's very original
00:09:29 <LordAro> Samu: is fine either way
00:09:37 <peter1138> Nice tremolo attect.
00:09:42 <LordAro> Samu: you do have to submit the review though
00:09:50 <peter1138> I only had one pint, I don't think it's alcohol making me typo.
00:10:16 <Eddi|zuHause> well, "ad libitum" means "do what you want" :p
00:10:21 <LordAro> nielsm: well it's not complete garbage :p
00:10:47 <peter1138> Little bit late after it's been merged.
00:11:15 <LordAro> Samu: commenting on merged PRs is rarely worthwhile
00:11:31 <LordAro> possibly better off making an issue
00:11:31 <peter1138> Open an issue referencing the PR.
00:12:33 <nielsm> pushed to my adlib-music branch :3
00:12:43 <nielsm> and now I see the time is past midnight
00:12:45 <peter1138> push early, push often
00:12:48 <peter1138> It's not, it's 23:12
00:13:10 <nielsm> it's later than I ought to be in bed
00:13:33 <peter1138> Pfft, have a cup of tea. It's your witching hour, where good ideas come to fruition.
00:13:47 <peter1138> I'm cheating, I have fennel tea.
00:14:15 <LordAro> oh hey, still some tea in here
00:14:36 <peter1138> I must remember to not actually drink caffeine after, say, 2 or 3pm.
00:14:52 <peter1138> Weirdly, stone cold tea isn't too bad to drink *when you're expecting it*
00:15:15 <LordAro> that's probably how i'd describe it :p
00:15:26 <LordAro> i have a tendency to forget my tea
00:15:47 <Eddi|zuHause> call it "ice tea"?
00:16:20 <peter1138> Iced tea with... milk. Hmm...
00:16:31 <peter1138> LordAro, same. Especially as I have a bucket mugg too.
00:16:44 <Samu> sorry, i'm terrible at using the website
00:17:06 <LordAro> that issue doesn't say anything at all
00:17:08 * peter1138 ponders replying to #7062, or playing Minecraft.
00:17:13 <LordAro> and relies on context that's only here
00:17:33 <peter1138> Wow, yeah, that's terrible.
00:17:53 <LordAro> Samu: the detail you use when responding to comments on your PRs is excellent, use that
00:17:58 <Samu> i dont know how to link stuff with each other
00:18:09 <peter1138> No need to link. Just write text.
00:18:28 <peter1138> Where were you trying to do. What happened. What did you expect?
00:19:00 <peter1138> Oh it's raining in Minecraft :/
00:19:20 <peter1138> And your title is bogus.
00:19:35 <peter1138> It should say what is happening now, not what wasn't happening in some older version.
00:20:17 <Eddi|zuHause> for some reason when it's raining in minecraft i have the urge to go fishing
00:20:34 <peter1138> You realise I literally just picked up my fishing rod...
00:21:07 <Eddi|zuHause> i pulled a crazy good bow out with the fishing rod
00:21:16 <Samu> i'm blocked, can't think properly
00:21:47 <Eddi|zuHause> and after updating beyond 1.8 i could even repair it
00:22:21 <peter1138> I'm always on latest release.
00:23:00 <LordAro> still needs more detail
00:23:02 <Eddi|zuHause> well, that was the latest release back then
00:23:10 <Eddi|zuHause> i don't play often
00:25:01 <Samu> i think im gonna copy paste irc text
00:25:17 <LordAro> Samu: pictures are more helpful than savegames as well
00:27:44 <LordAro> not to say that the savegame isn't useful
00:28:06 <peter1138> Damn it, I keep picking up... fish.
00:28:21 <peter1138> Oh and two water bottles. Nothing interesting :(
00:32:18 <Samu> some of the AIs in this situation would mass even more ships
00:35:19 <Samu> because there's more cargo waiting at stations
00:35:57 <Samu> when the ship comes back, it reinvokes the pathfinder
00:37:33 <Samu> thx for the edit LardAro
00:39:52 <Samu> the only solution is path cache
00:40:46 <Samu> a means to avoid pathfinder being called every tile
00:42:43 <LordAro> as well as removing the manhattan check, i can see that working
00:42:50 <LordAro> caching the paths would improve things generally
00:44:50 <peter1138> The issue is actually that the orders should be invalid.
00:44:57 <peter1138> Buoys are necessary.
00:45:41 <peter1138> 22:46 < Samu> have the pathfinder invoked on order insertion
00:46:10 <peter1138> But not if the map is changed later.
00:46:37 <peter1138> The ship is lost because its destination is too far. You need buoys.
00:47:31 <LordAro> peter1138: but if the order was valid beforehand...
00:47:42 <LordAro> it would be a shame to break saves like that
00:47:54 <peter1138> It's not valid, it just worked because we didn't check properly.
00:48:27 <peter1138> Of course you can remove that check, but then you get performance hits again.
00:48:42 * LordAro is reminded of a Linus "never break userland" rant
00:49:25 <Samu> it does comes back into range only to have the pathfinder make him lost again
00:51:13 <peter1138> Were always necessary in this situation
00:51:29 <Samu> nop, it's not lost in 1.8.0
00:51:54 <LordAro> peter1138: well, they were only necessary because of the bad pathfinder
00:52:36 <Samu> i think the savegame is 1.8.0 compatible
00:53:22 <peter1138> LordAro, of course, in this particular map the pathfinder wouldn't have too much trouble.
00:53:34 <peter1138> Open spaces of water is the issue.
00:54:09 <LordAro> how hard would it be to make the pathfinder prefer the shoreline?
00:54:43 <Eddi|zuHause> how would that help anyone?
00:54:54 <Samu> with costs, it's doable, somehow
00:55:15 <LordAro> Eddi|zuHause: less open spaces of water
00:55:24 <Samu> but then it would have to detect coast tiles
00:55:34 <Eddi|zuHause> that doesn't reduce the search space
00:55:37 <Samu> do more checks every tile
00:55:56 <Eddi|zuHause> just make it find worse paths
00:58:21 <Eddi|zuHause> to reduce the search space across open water, you could try to merge tiles into 2^n square superblocks that contain no obstacles
00:59:02 <Eddi|zuHause> store the size of the superblock in the tile
00:59:19 <Eddi|zuHause> upon changing water (clearing, flooding), invalidate the nearby superblocks, and recalculate
01:00:10 <peter1138> There is m8 now, heh.
01:00:13 <Eddi|zuHause> superblocks are also 2^n-aligned
01:00:22 <peter1138> But not much use for narrow paths, such as canals.
01:00:31 <Eddi|zuHause> canals are not the problem
01:02:05 <peter1138> I can probably do the superblock marks, but no idea how to plug that into the pathfinder.
01:02:24 <LordAro> press buttons until it works
01:02:42 <Eddi|zuHause> a superblock is just a more complicated follow_track function, as you have more entry and exit points
01:03:06 <peter1138> Also your superblocks could potentially have dozens of exits and entrances.
01:03:24 <peter1138> And the cost would differ depending on where you start.
01:03:26 <Eddi|zuHause> but all the internal paths can be ignored
01:03:50 <Eddi|zuHause> you can calculate the cost based on entry and exit direction
01:04:00 <Eddi|zuHause> because you know all the tiles inbetween are empty
01:04:55 <Samu> could pathfinding be multi-threaded?
01:05:14 <Eddi|zuHause> you need some idea how to combine all the touching tiles between two superblocks
01:07:22 <Eddi|zuHause> but that could be an optimisation on top, it probably is already an improvement even if you consider every possible connection between 2 superblocks individually
01:07:56 <Eddi|zuHause> sidenote: i don't think you need m8, water should have plenty of free bits
01:08:41 <peter1138> True, it's only clear water.
01:11:41 <peter1138> Hmm, you could perhaps extend it to rectangular superblocks.
01:12:07 <Eddi|zuHause> yes, but that makes checking/storing it more tricky
01:12:09 <peter1138> Although that complicates path lengths I guess.
01:12:33 <peter1138> Still 2^ aligned, just separate x/y.
01:13:18 <Eddi|zuHause> but it might not help much, and you have potential conflicts whether to extend in x or y direction
01:25:07 <Samu> gonna experiment making it pathfind every other tile
01:26:44 <Samu> (DistanceManhattan(tile, v->dest_tile) > SHIP_MAX_ORDER_DISTANCE + 5) & 1
01:32:36 <Samu> if (v->dest_tile == 0 || DistanceManhattan(tile, v->dest_tile) > SHIP_MAX_ORDER_DISTANCE + 5 && DistanceManhattan(tile, v->dest_tile) & 1) {
01:33:10 <peter1138> NPF works properly. It can't find a path between the two docks :p
01:37:54 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on issue #7062: Ship becomes lost if destination is greater than maximum order distance https://git.io/fhcVy
01:37:59 <Samu> oh lol, ship is lost... no, nop, it's not lost, wait, it's lost again
01:38:38 <peter1138> If you actually only want to pathfind less frequently, you can do that a better way.
01:38:47 <peter1138> (DistanceManhattan not involved)
01:39:18 <Samu> doesn't work well, just tested, ship still doesn't go to where it should
01:40:59 <peter1138> If you don't pathfind, you have to assume a direction to continue in.
01:41:08 <peter1138> (Or, indeed, cache the path)
01:43:28 *** Thedarkb-T60 has joined #openttd
01:44:00 *** Smedles_ has joined #openttd
01:50:19 <Samu> (DistanceManhattan(tile, v->dest_tile) & 3) == 0
01:50:31 <Samu> ship still does weird stuff at times but...
01:50:35 <peter1138> You're barking up the wrong tree.
01:54:36 <Samu> if (v->dest_tile == 0 || DistanceManhattan(tile, v->dest_tile) > SHIP_MAX_ORDER_DISTANCE + 5 && Random() & 1) {
01:56:07 <Samu> it works, but yeah, it's wrong lol
01:56:22 <Samu> never know how random is random
01:57:40 <Samu> rellying on random to improve pathfinder speed
02:02:01 <Samu> this might actually be a "good enough" solution? let me try the 5000 ship savegame
02:03:03 <Samu> i suspect it too be either too spiky or too smooth
02:07:43 <Samu> hmm the ships look rather dumb
02:08:01 <Samu> doesn't look that good to the eye
02:11:46 <Samu> 5000 ships, on a 1024x1024 map with lots of water, fast forwarding at 52 fps, not too bad imo
02:12:21 <Samu> quite irregular, ranges 50-70
02:16:15 <Samu> testing without randomness
02:20:04 <Samu> getting the same fps which is strage
02:25:29 <Samu> i think it may be worth trying the && Random() & 1
02:25:36 <Samu> after the distance check
02:26:19 <Samu> track choice when it fails could be improved somewhat
02:26:45 <peter1138> It continues in the current direction.
02:27:44 *** Thedarkb-X40 has joined #openttd
02:28:25 <Samu> if it goes vertical or horizontal against a coast, it could make a 45 degree turn
02:28:49 <Samu> instead of being forced to reverse
02:29:37 <Samu> depends on coast direction too I suppose
02:32:05 <Samu> i'm off to bed, take care
02:35:02 <DorpsGek_II> [OpenTTD/OpenTTD] glx22 opened pull request #7063: Fix: deps calculation call could fail due to command line length https://git.io/fhcKl
06:04:28 *** FLHerne has joined #openttd
07:43:42 *** andythenorth has joined #openttd
08:11:03 *** andythenorth has joined #openttd
08:14:36 *** chomwitt has joined #openttd
08:51:51 *** AKTheKnight has joined #openttd
08:54:37 *** D-HUND is now known as debdog
09:04:48 *** WWacko1976-work has joined #openttd
09:06:21 *** andythenorth has joined #openttd
09:32:27 *** Thedarkb-T60 has joined #openttd
09:55:11 *** andythenorth has joined #openttd
10:14:03 <DorpsGek_II> [OpenTTD/OpenTTD] Gabda87 commented on pull request #7047: Add #6887: Highlight tiles within local authority of towns https://git.io/fhcxF
10:21:32 *** andythenorth has joined #openttd
11:03:03 <peter1138> Why is my tummy rumbling? I had breakfast :S
11:43:56 *** sla_ro|master has joined #openttd
13:26:34 <DorpsGek_II> [OpenTTD/OpenTTD] Gabda87 opened pull request #7064: Fix #6438: raising/lowering 'Maximum no. competitors' setting in multiplayer https://git.io/fhCGC
13:34:20 *** raphikov has joined #openttd
13:46:45 <andythenorth> peter1138: lunch time yet?
14:05:24 *** snail_UES_ has joined #openttd
14:17:31 *** ChanServ sets mode: +v tokai
15:22:40 <peter1138> It was, I'd already eaten.
15:22:48 <peter1138> But only about 5-10 minutes before-hand.
15:26:34 <Samu> mirc expires after 30 days?
15:27:16 <peter1138> But mIRC is pretty shit.
15:33:18 <Samu> how to persuade him into looking my code?
15:36:40 <andythenorth> he'll either comment on your PR because it looks interesting, or he won't
16:00:21 <peter1138> Should I investigate super-tiles?
16:01:19 <nielsm> as in splitting the map up in blocks of tiles?
16:01:34 <peter1138> Not actually splitting.
16:01:37 <nielsm> macro-tiles, to make a comparison to video coding
16:01:54 <peter1138> Just recording such for open water.
16:02:44 <andythenorth> is this for pathfinding?
16:02:54 <nielsm> yes some kind of optimization for determining water connectivity is likely good
16:03:07 <andythenorth> does it have any relevance to also giving towns blocks of super tile?
16:03:27 *** WWacko1976-work has left #openttd
16:03:28 * andythenorth deletes all the ships
16:15:04 <peter1138> Maybe I should just test my existing caching patch a bit more. I don't think super-tiles will solve the unreachable tile issue.
16:15:45 <Eddi|zuHause> unreachable will always be the worst case
16:16:03 <peter1138> It's such a bad worst case though :)
16:16:24 <Eddi|zuHause> you can make region numbers
16:16:26 <peter1138> Ah well, at least you are less likely to have 5000 ships all concurrently doing it now.
16:16:47 <peter1138> Actually m2 is free for water tiles anyway.
16:17:37 <Eddi|zuHause> problem with region numbers is, in case of clearing a tile, you must look whether all adjacent tiles are still connected
16:17:39 <peter1138> I wonder how to calculate regions efficiently.
16:17:51 <Eddi|zuHause> and potentially allocate a new region
16:17:58 <Eddi|zuHause> and on flooding you might have to merge regions
16:18:17 <Eddi|zuHause> otherwise, it's just a flood fill
16:18:21 <peter1138> Merging is simpler than splitting, I think.
16:19:14 <Eddi|zuHause> finding tiles that are not yet in any region is a tricky part as well
16:19:19 <peter1138> Yes really. With merging you have just 1 tile that connects, you can simply mark all the other tiles in that region to the new merged region.
16:19:54 <peter1138> But for splitting you need to keep track of way more stuff.
16:20:02 <peter1138> And indeed finding the regions, yes.
16:20:18 <Eddi|zuHause> with splitting the only tricky part is to detect whether it's still connected
16:20:36 <Eddi|zuHause> once you detected that, it's again just a flood-fill
16:20:42 <andythenorth> it's really quite a nice problem
16:20:52 <peter1138> andythenorth, that sounds like volunteering :D
16:21:07 <andythenorth> but it's nicely non-trivial
16:21:16 <andythenorth> and also makes piss all difference in most games :)
16:21:34 <Eddi|zuHause> the word you were looking for is "niche" :p
16:22:21 <peter1138> Alternatively you can finite-size regions, and maintain a list of connected regions.
16:22:55 <peter1138> Although that complicates the "can you get from here to there" test.
16:23:04 <Eddi|zuHause> someone did that with placing a bunch of buoys in the center of each region
16:34:33 <peter1138> Mint Daim bars are pretty damn good. Shame they're all processed sugar :/
16:39:55 <Eddi|zuHause> processed sugar is not worse than natural sugar
16:40:47 <peter1138> Actually the form-factor of your carbohydrates does affect things.
16:41:15 <peter1138> But yeah. Shame they're all sugar. :p
16:41:51 <Eddi|zuHause> but the most important thing is the mixture. once the sugar is removed from the original fruit (like, say, fruit juice), you've basically lost
16:42:41 <peter1138> Fruit juice is basically concentrated sugar. I haven't bought any for ages.
16:45:15 <peter1138> Over Christmas I bought a load of goodies in lots of moments of weakness :p
16:45:22 *** Wormnest has joined #openttd
16:45:52 <peter1138> Working my way through them means I've got high sugar for a bit. I could just bin them but I hate food waste.
16:48:07 <Eddi|zuHause> oh, and the worst part is when you mix sugar and proteins
16:48:45 <andythenorth> sounds like a cheese toasty with jam to me
16:48:59 <peter1138> No need for the jam.
16:50:45 <peter1138> Plenty of carbs in bread, though the sugar is relatively lower.
16:53:07 <peter1138> Still, the carbs gets turned into sugar anyway.
16:57:08 <nielsm> now it's making the sound actually correct!
16:58:06 <peter1138> Hmm, sounds broken to me. Maybe that's my sound.
16:58:17 *** Wormnest_ has joined #openttd
16:58:53 <peter1138> Yeah it's not right.
17:00:11 <nielsm> it's wrong, but it's not chopped up
17:00:20 *** Thedarkb-T60 has joined #openttd
17:00:22 <peter1138> It is choppy for me. Hmm.
17:00:46 <nielsm> before I was mishandling the PCM buffer so half the buffer was filled with zeroes and the other half mixed-up samples
17:02:25 <nielsm> the remaining bugs should be traceable to actual faults in my reimplementation of the decoder, as opposed to dumb mistakes in pcm buffer handling
17:02:42 <nielsm> (wrote 2*n samples to the buffer and incremented pointer by n samples)
17:30:38 <nielsm> gah okay, I'm doing _something_ bad here, because I'm hitting heap corruption triggers whenever my new code has been running
17:35:58 * peter1138 rides out the sugar crash.
17:36:17 * andythenorth wants a sugar crash
17:36:21 * andythenorth had a no-sugar crash :(
17:40:53 <AKTheKnight> Is it a day at work if I don't have a crash at some point?
17:41:26 <Samu> i'm experimenting different costs for water stuff
17:41:29 <peter1138> It's a ssensible day.
17:41:55 *** Thedarkb-T60 has joined #openttd
17:42:16 <Samu> if IsWaterTile, add a big cost, so that in essence i'm making the other water tracks cheaper for the pathfinder
17:42:42 <Samu> because they're in less quantity, i wonder if this helps pathfinding along the coast
17:43:26 <peter1138> Actually I'm probably just tired from lack of sleep, and just sat here no moving :p
17:43:52 <Samu> pathfind in less node searchs
17:44:35 *** daspork has joined #openttd
17:45:41 <nielsm> and still triggers heap corruption
17:47:41 <AKTheKnight> Can you add that as an actual soundtrack? I kinda like it
17:48:07 <Samu> costs are important after all
17:48:18 <Samu> what was a valid path, now became a lost path
17:48:31 <nielsm> the bugs when I was working on the GM decoder for dos music were more interesting :)
17:55:13 <peter1138> nielsm, lol, after the intro worked so perfectly ...
17:55:21 *** Progman has joined #openttd
18:03:49 <Samu> just tested no penalty for making turns, and it appears yapf does find paths slightly faster, 1 test case though
18:04:00 <Samu> ships dont slow down on turns
18:05:04 <Samu> gonna test with 5000 ships, brb
18:05:49 <Samu> oh wait, this is debug mode, crap
18:14:50 <Samu> meh, with 5000 ships the difference is negligible, not worth the effort
18:15:33 <Samu> penalty for curves apparently does help 1 or 2%
18:26:09 <Samu> holy crap beguglevel npf=9 is suicide
18:32:26 <peter1138> It's a lot of useful debug information.
18:32:32 <peter1138> You probably don't want to use it with 5000 ships.
18:33:33 <peter1138> I love how you just get random ideas and keep plugging away with different variations without really knowing what you're doing :p
18:34:06 <andythenorth> peter1138: I assume you're talking to me ^ :P
18:34:19 <andythenorth> sounds like UI design
18:34:33 <andythenorth> and my entire career
18:34:47 <andythenorth> very much children
18:35:04 <andythenorth> so I've coded EMUs that are slow-ish
18:35:09 <andythenorth> unless you attach them to an engine
18:35:14 <andythenorth> then they're fast-ish
18:35:19 <andythenorth> worst BAD FEATURE so far?
18:35:56 <andythenorth> peak BAD FEATURE
18:37:09 <Samu> horizontal tracks cost 71 on yapf, 70 on npf, is that intended?
18:37:36 <LordAro> Samu: probably not, but i don't expect it to make any significant difference
18:37:58 <andythenorth> it's like realism, if I controlled reality
18:38:05 <andythenorth> it's not at all like realism otherwise :P
18:38:11 <andythenorth> I might retcon it somehow though
18:38:39 <andythenorth> just amuses me to hide stupid features :D
18:38:41 <peter1138> Damn, Continuum shaders are kinda slow.
18:38:53 *** daspork has joined #openttd
18:39:46 <Samu> sqrt(2)/2*NPF_TILE_LENGTH =~ 0.7071
18:40:24 <Samu> #define STRAIGHT_TRACK_LENGTH 7071/10000
18:40:36 <Samu> yapf doesn't do any of this math
18:43:17 <andythenorth> should I also make it change the speed
18:43:26 <andythenorth> if the engine is at the rear of the consist?
18:45:05 *** HerzogDeXtEr has joined #openttd
18:49:05 <andythenorth> don't think I have a var for that :P
18:50:32 <andythenorth> requires pissing around with var 61
18:53:46 <Eddi|zuHause> just check the position in consist in the engine?
18:53:58 <Eddi|zuHause> (works only for lowering speed, not raising)
18:54:14 <Eddi|zuHause> but i still don't think you're going in any sane direction
18:54:19 <andythenorth> I'm seeing if I can count IDs
18:54:25 <andythenorth> without too much nml
18:56:02 <Eddi|zuHause> there's a var for that i think
18:58:16 <Samu> the real time it takes to walk a tile in axis direction is 16*256* advance speed i think, gonna recheck
19:05:41 *** gelignite has joined #openttd
19:09:59 * andythenorth wonders what the worst feature I could add is?
19:10:16 <andythenorth> snail has implemented brake force in french NG set, that's quite interesting
19:12:09 <Eddi|zuHause> in some sense snail is even worse than george :p
19:12:59 <Samu> oops i forked andythenorth notwater stuff by mistake
19:13:04 <andythenorth> it's comprehensive and detailed and works though
19:13:17 <andythenorth> there is a clear theme to it, and a big user guide
19:16:11 <Samu> wanted to head to my own fork, and instead it did that :(
19:18:07 *** Mahjong has joined #openttd
19:21:28 <Samu> The vehicle speed is scaled by 3/4 when not moving in a DiagDir direction due to the shorter distance I think would be correct
19:22:40 <Samu> or i don't know the meaning of scaled by
19:35:20 <andythenorth> I could add different types of brakes?
19:35:46 <andythenorth> compatibility rules for multi-head engines?
19:35:58 <andythenorth> pikka already did pax train electric supply :P
19:36:38 <Samu> a tile has 16 sublocations, and needs 256 progress units to advance 1 sublocation
19:36:57 <Samu> and vehicle moves at full speed
19:37:05 <Eddi|zuHause> pikka had this minimalist phase where EVERY feature is a BAD FEATURE
19:37:25 <Samu> another tile, has 8 sublocations, and needs 256 progress units to advance 1 sublocation, and the vehicle moves at 3/4 speed
19:37:37 <Eddi|zuHause> "only 10 engines" and such
19:37:38 <andythenorth> pikka need to reset :)
19:37:55 <Samu> for pathfinder cost penalties, if the first one costs 100, what would the other cost?
19:38:09 <andythenorth> so should I do variable running costs for engines, relative to powered-ness on current railtype?
19:40:09 <andythenorth> might need a var for 'powered-ness' :P
19:41:30 <Samu> i'm more inclined towards 75 because it makes more sense in my mind, lol
19:41:37 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z updated pull request #7000: Add: Var 6A, a clone of Var 4A for querying poweredness compared to a… https://git.io/fhI7h
19:44:02 <Wolf01> <Samu> I need a math expert :| -> @calc?
19:46:22 <Eddi|zuHause> Wolf01: calculation has nothing to do with maths
19:49:29 <Eddi|zuHause> aw, i forgot git add :/
19:49:42 <Eddi|zuHause> how do i add just one changeset in a file?
19:50:33 <LordAro> Oh, a single changeset: -p
19:52:14 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z updated pull request #7000: Add: Var 6A, a clone of Var 4A for querying poweredness compared to a… https://git.io/fhI7h
19:54:06 <Samu> the length of a straight track is 50
19:54:20 <Samu> and the vehicle moves at 3/4
19:54:39 <Samu> cant use decimals, what can i do
19:55:34 <Eddi|zuHause> someone should look over my bitmath and tell me whether i got that right
19:58:34 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z updated pull request #7000: Add: Var 6A, a clone of Var 4A for querying poweredness compared to a… https://git.io/fhI7h
19:59:07 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z updated pull request #7000: Add: Var 6A, a clone of Var 4A for querying poweredness compared to a… https://git.io/fhI7h
20:04:43 <Samu> i just made opf ships drunk
20:05:11 <andythenorth> Eddi|zuHause: you can issue sparks from somewhere else :P
20:05:39 <Samu> it looks the answer is not 37.5, or i'm not sure i'm visualizing this right
20:06:12 *** frosch123 has joined #openttd
20:06:33 <andythenorth> maybe I'm wrong, maybe effect vehicles aren't in tthe spec :)
20:07:18 <Samu> a diag track has 16 sub locations, a vehicle needs 256 progress units to advance a sub location
20:08:00 <Samu> and vehicles move at 3/4 on them
20:08:26 <Samu> which is like saying it needs 256*4/3 ?
20:10:12 <Eddi|zuHause> andythenorth: you can move them forwards and backwards
20:10:20 <Eddi|zuHause> but not up and down, or sideways
20:10:33 * andythenorth makes peace with that
20:10:50 <andythenorth> sure I've seen effect vehicles spec somewhere in frosch123 wiki :)
20:10:52 <Eddi|zuHause> plus if you want to make 3rd rail sparks, you also need to know which side the 3rd rail is on
20:11:29 <andythenorth> I tried to add 3rd rail flash to 00 trains once
20:11:39 <andythenorth> with an LED, a reed switch and a magnet
20:13:01 <Eddi|zuHause> frosch123: when andy says "frosch wiki" i'm immediately assuming it's not implemented :p
20:13:46 <Eddi|zuHause> about the railtype speed, i'm somehow questioning the usefulness if that, since you don't know whether that's the currently limiting speed
20:13:48 <frosch123> rude, about 1/4 is implemennted
20:13:52 <andythenorth> the frosch wiki is a happy place
20:13:57 <andythenorth> things go there to ripen
20:14:11 <andythenorth> if yexo was here, more of it would be picked
20:14:26 <Eddi|zuHause> frosch123: but the implemented parts should also be on other wikis :p
20:14:48 <andythenorth> how do we get a new yexo?
20:15:56 <andythenorth> I mean it's possible, in theory
20:18:20 <Eddi|zuHause> uhm... need to rethink these last commits
20:26:34 <Eddi|zuHause> i think i took a wrong turn somewhere
20:28:32 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z updated pull request #7000: Add: Var 6A, a clone of Var 4A for querying poweredness compared to a… https://git.io/fhI7h
20:32:17 <Samu> if npf is similar to yapf, why does it require about 25000 nodes to find what yapf finds in 10000?
20:33:46 <andythenorth> why did yapf 'replace' npf? o_O
20:33:55 <peter1138> Optimisations affect how it works.
20:34:45 <Eddi|zuHause> yapf was written by a crazyperson
20:34:59 <frosch123> npf visits every tile whenever searching
20:35:11 <frosch123> yapf caches segments and their cost between junctions
20:35:31 <Samu> for ships there is no cache, if i'm not mistaken
20:36:07 <frosch123> which works well if you have more track than junctions
20:37:30 <LordAro> so turning large sections of open water into "segments" would work well then!
20:38:29 <Eddi|zuHause> LordAro: that's where the supertiles would come in
20:38:43 <Eddi|zuHause> not quite the same thing, though
20:39:02 <frosch123> ships have two issues
20:39:32 <frosch123> 1. there are junctions everywhere, so pathfinder has lots of options. 2. there are junctions everywhere, so the ship has to redecide on every tile
20:39:37 <Samu> yapf is better for ships, but I don't know why, can't figure the reason myself
20:40:07 <frosch123> somewhen i tried to address (2), by remembering how many tiles a ship won't change direction before searching next time
20:40:15 <frosch123> but did not finish :p
20:40:24 <Eddi|zuHause> was just about to ask how far that got :p
20:41:03 <peter1138> Why do you think there is no cache?
20:41:20 <Samu> there is a cache that always returns false
20:41:28 <Samu> so it's not enabled where i tested?
20:43:21 *** Thedarkb-T60 has joined #openttd
20:43:47 <Samu> seems to be a thing for roads and rails using pfs if I recall
20:44:05 <Samu> caches the entire road segment until it finds a turn
20:44:24 <Samu> for ships... i can't see anything
20:45:20 <andythenorth> do ships really pathfind on every tile? :o
20:45:25 <andythenorth> instead of caching a vector?
20:57:08 <andythenorth> what is the problem with ships?
20:57:34 <Eddi|zuHause> there is no problem with ships, everything works fine
20:57:58 <andythenorth> why do they need all these changes then? :P
20:58:07 <andythenorth> I just use ships in my game
20:58:12 <andythenorth> and if I use it a lot, it gets slow
20:58:24 <nielsm> hmm in its current broken form, the adlib player takes about 3 ms to produce 8192 samples at 44.1 khz
20:59:34 <glx> and 8192@44.1 time lenght is ?
21:00:32 <nielsm> so about 16 ms out of every second is spent on music?
21:00:40 <nielsm> (which runs on a separate thread)
21:01:20 <peter1138> There's a reason these things were done in hardware back in the 90s.
21:01:50 <nielsm> and besides the hardware synth is analog signal generators with digital controls
21:02:21 <glx> it should be possible to prepare a sample batch while another is actually played
21:02:54 <nielsm> that just requires some synchronization I'm not into doing right now
21:03:16 <nielsm> wow something breaks completely in sawyer's tune
21:03:23 <peter1138> That shouldn't be necessary.
21:03:24 <nielsm> a few places it hangs weirdly
21:03:42 <peter1138> The audio is already buffered, so it already is being played concurrently with generation.
21:03:45 <glx> fill the buffer, wait for mixer to ask for the buffer
21:03:59 <glx> give it and prepare the next batch
21:04:37 <glx> you'll still need to copy your buffer into the mixer provided buffer
21:25:34 <Eddi|zuHause> i think this is how far i can sensibly do it righ now, the parts about what's the speed limit and what's the cruising state is a bit more elaborate
21:26:24 <andythenorth> I missed those proposals in the comments
21:26:28 <andythenorth> but they're very interesting
21:27:14 <Eddi|zuHause> still unsure about the speed limit thing in var 4A
21:28:28 <DorpsGek_II> [OpenTTD/OpenTTD] andythenorth commented on issue #6947: A new variable Vehicle is loading required https://git.io/fhCj1
21:29:15 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z commented on pull request #7000: Some NewGRF variables concerning railtypes https://git.io/fhCjH
21:30:56 <Eddi|zuHause> speed limit thing might be something for GroundVehicle rather than Train?
21:31:20 <Eddi|zuHause> planes could also have speed limit while taxiing
21:31:26 <andythenorth> I could make a dual-mode 3rd-rail and overhead AC loco? :P
21:32:17 <andythenorth> planes with a pushback tug :P
21:33:09 <andythenorth> not sure what I'd do with the speed limit var, but I'm not george or snail et al
21:37:28 <Eddi|zuHause> i'm thinking a "speed_limit" variable, and a "speed_limit_flags" variable in Vehicle, and the acceleration/movement code sets this appropriately. on every line that sets the speed_limit variable, it is compared with the current speed_limit value, and if lower, speed_limit_flags is cleared and new bit is set (depeding on line), if equal, lime-bit is added to the current flags, and if higher, nothing is done
21:38:17 <peter1138> Hmm, right, maybe I should work on some things I promised, or have already started on.
21:38:33 <peter1138> I wonder why gitg removed its stash viewer. That was really useful.
21:38:33 <Eddi|zuHause> "speed_limit_flags" might be "vehicle", "track", "bridge", "station approach", "timetable", ...
21:39:08 <Eddi|zuHause> "power exhausted"?
21:39:19 <Eddi|zuHause> (i.e. acceleration was 0)
21:39:57 <Eddi|zuHause> or "te exhausted"?
21:41:13 <andythenorth> peter1138: work on new things!
21:41:33 <peter1138> Hmm, ships stopping in locks...?
21:41:35 <andythenorth> no need for docs and making the test pas
21:41:39 <peter1138> That's about 10 years old.
21:41:39 <andythenorth> and all that crap
21:41:49 <Eddi|zuHause> peter1138: newgrf statemachines
21:42:05 <peter1138> Not sure that was something I ever planned, Eddi|zuHause;
21:42:07 <andythenorth> newgrf station tiles that can show more than one cargo (across the whole station)
21:42:21 <Eddi|zuHause> (would cover ship locks, drawbridges, highways/onramps, airports)
21:42:39 <andythenorth> flat docks are allegedly in the spec already, mumble mumble TTDP
21:42:41 <Eddi|zuHause> (tram turning loops)
21:43:03 <andythenorth> unrelated: anyone know how physics works?
21:43:39 * andythenorth wonders about an acceleration factor, it's a proxy for gearing, peak torque etc
21:43:42 <andythenorth> TE isn't the same
21:44:05 <andythenorth> TE has almost no effect unless you add enough weight to a train that it's on it's knees on a slope
21:45:02 <andythenorth> the significant property for train performance in game is purely HP, except for pathological combination of weight + slope
21:45:13 <Eddi|zuHause> train needs a startup force? won't move unless TE > that force?
21:45:31 <andythenorth> yes, but if TE is 'enough' there isn't much more observable difference
21:45:42 <andythenorth> I did learn how TE works once, and I believe that's correct
21:46:03 <andythenorth> whereas acceleration depends on the rate of work or so
21:46:18 <peter1138> problem is we use the max TE all the time, which in reality would result in an... uncomfortable and inefficient acceleration.
21:46:24 <Eddi|zuHause> TE is the limiting factor on very low speeds (<20km/h or so)
21:46:37 <andythenorth> for acceleration, mechanical or electrical factors like gearing, wheel size, or max watts on the motor are the factors
21:46:50 <andythenorth> TL;DR I'd like to be able to make some trains suited for many stops
21:46:55 <andythenorth> and some for few stops
21:47:03 <andythenorth> it's hardly necessary
21:47:13 <peter1138> basically the only way to have slow acceleration is to have very underpowered trains
21:47:15 <andythenorth> but currently I have to piss around with quite fine adjustments of HP
21:47:45 <andythenorth> fast trains that have enough HP to achieve max speed are quite OP for all other cases
21:47:49 <Eddi|zuHause> other than braking force, i don't see what you can really change there
21:48:01 <andythenorth> I read the code, and I didn't see an obvious thing to try
21:48:08 <Eddi|zuHause> or ridiculously high running costs
21:48:16 <andythenorth> but there's probably a rate of acceleraion that could have a factor applied
21:49:07 <Eddi|zuHause> "real" trains are also probably scheduled about 80% of their max speed
21:49:25 <andythenorth> acceleration out of stops is a key issue IRL
21:49:29 <andythenorth> for pathing and performance
21:49:34 <andythenorth> not sure that matters in game
21:50:00 <Eddi|zuHause> you don't have enough controls to handle that properly
21:50:17 <andythenorth> but in Horse, unless you load the weight up enough to put the train on its knees, you are usually better off choosing the expensive pax engine
21:50:38 <andythenorth> it will beat the equivalent sized freight engine over the same tile route
21:50:48 <andythenorth> it will cost more, but there's less network contention
21:50:54 <andythenorth> and network contention >>>> money
21:51:04 <peter1138> Hmm, should I make this patch a setting? :p
21:51:16 <andythenorth> accelerating out of signal stops is quite a big deal in the kind of crap ottd networks I build
21:51:21 * andythenorth is not coop style
21:51:22 <Eddi|zuHause> settings are evil
21:51:39 <peter1138> Yes but gameplay changes are ... hmm ...
21:51:58 <andythenorth> what changed? :P
21:52:06 <peter1138> I considered that but it's too much work.
21:52:29 <peter1138> I just resigned from being the local CTC ride coordinator because I'm too lazy to do it once a month.
21:52:48 <peter1138> andythenorth, this is my patch that makes ships turn slowly instead of instantly.
21:53:30 <peter1138> It's quite an invasive patch for something that isn't seen very often :/
21:53:41 <peter1138> Maybe we need tugs :p
21:55:05 <peter1138> ^ It's that last change.
21:56:07 <andythenorth> do it anyway? o_O
21:56:12 <andythenorth> forums always blame me
21:56:20 <andythenorth> or someone innocent, never you
21:56:28 <peter1138> You hate it? Did you... what?
21:56:48 <andythenorth> of course I didn't
21:57:23 <andythenorth> do I have you upstream?
21:58:17 <andythenorth> hmm where is acceleration stuff, ground_vehicle_cmd maybe?
22:00:29 <peter1138> Fitbit says I can eat another 1000 kcal, despite have a pizza and garlic bread for dinner. Madness. Is cheese a good idea?
22:01:19 <andythenorth> peter1138: how do I test the reversing?
22:01:59 <peter1138> Basically when ship heads into a dock head on, when it leaves it will turn instead of just flipping around.
22:02:12 <andythenorth> and if I reverse it en-route (change order)?
22:02:19 <peter1138> Now I think about it, this is massively over engineered for something that is barely visible :s
22:02:24 <peter1138> Yeah that should too.
22:02:41 <andythenorth> no actually I love it
22:02:50 <peter1138> And 90 degree turns will have an intermediate step.
22:02:51 <andythenorth> it's a lot more like real ships
22:02:56 <andythenorth> it's a bit ponderous
22:03:07 <peter1138> It makes ships worse, just what we need :D
22:03:10 <andythenorth> I thought it might be annoying in a game, but it's a small-big-improvement
22:03:32 <andythenorth> looks like they're maneouvering
22:04:01 <andythenorth> yeah that 180 deg dock flip :P
22:04:03 <andythenorth> I won't miss that
22:04:25 *** Maarten has joined #openttd
22:05:06 <andythenorth> hmm if we have air drag
22:05:19 <andythenorth> we can probably accomodate a proxy for gearing / etc
22:05:26 <andythenorth> dunno what the technical term would be
22:05:35 <andythenorth> is it some kind of torque co-efficient?
22:06:00 <andythenorth> it applies during acceleration, rather than cruise
22:08:03 <peter1138> just have a callback, called every tick, to have the vehicle calculate its own acceleration :p
22:08:55 <andythenorth> I threw out more mouldy cheese
22:08:58 <peter1138> I got a big wedge of stilton now.
22:09:14 <AKTheKnight> Favourite cheese?
22:09:30 <andythenorth> I tried cheese in FIRS
22:09:36 <andythenorth> didn't quite work, but eh
22:09:40 <peter1138> AKTheKnight, all of it.
22:10:10 <AKTheKnight> Haha if you had to pick one?
22:11:24 <peter1138> Oh yeah I got a big wedge of mature Gouda too.
22:11:42 <andythenorth> newgrf can then do all the things like phase transition :P
22:11:47 <AKTheKnight> Oooh, I love Gouda
22:11:52 <andythenorth> efficient use of regulator / controller :P
22:11:55 <AKTheKnight> Mont d'Or has gotta be my all time favourite though
22:12:38 <peter1138> Baked Mont d'Or is good
22:12:56 <peter1138> Bloody hell, cheese is so good.
22:13:03 <peter1138> Why does it have to have so much unhealthy fat :(
22:13:47 <peter1138> Mind you kinda like Philidelphia too, but that's pretty bad cos it's not really cheese.
22:14:00 <AKTheKnight> Bagels, cream cheese, smoked salmon
22:14:03 <AKTheKnight> Perfect breakfast
22:14:35 <AKTheKnight> You trying to go low carbs?
22:14:49 <AKTheKnight> I really should care about what I eat, but for now I'm alright
22:15:39 <peter1138> Not currently, but I cut back on carbs a bit for a few months last year and felt a lot better. Lost a shed load of weight too, probably not a coincidence.
22:16:21 <peter1138> I want to cut back a bit now as I had fuck loads over christmas and they're kinda self-fulfilling. You want more while you're having them.
22:16:30 <AKTheKnight> That makes sense. I'm cutting back on takeaways/unhealthy and trying to lower my portion sizes a bit too
22:17:17 <peter1138> Carbs leads to be snacking, which results in overeating.
22:18:43 <AKTheKnight> Yeah I've cut back on snacking a bit, keeping myself busy at work helps with that
22:18:47 <peter1138> Although there's a balance, I became a bit too underweight which is also not great.
22:19:09 <peter1138> As a previously fat person who thought "bah, how can you be underweight, just eat" ... well, it's not that simple :)
22:19:46 <AKTheKnight> Yeah I'm trying to avoid putting on anymore weight and being the fat person
22:22:00 <peter1138> 3 years of solid cycling helped too!
22:23:20 <AKTheKnight> I need to get into doing more exercise. Think I'll try and do some running and cycling again once it warms up
22:26:01 <peter1138> Hmm, lots of HF noise?
22:26:13 <peter1138> or is that LFO gone mad?
22:26:23 <peter1138> Ah, just saw the time scale.
22:26:26 <peter1138> It's an LFO gone mad.
22:26:56 <nielsm> yes looks like around 0.25 hz
22:27:13 <peter1138> In an analogue system that DC offset would be lost due to output capacitors filtering it.
22:27:24 <peter1138> (And indeed, it will be)
22:27:59 <nielsm> mostly the percussion is bugged
22:28:17 <nielsm> and then something about the delays
22:29:21 <nielsm> though this is with the dual channel/dual track code disabled, since that causes a bunch of issues right now
22:30:47 <peter1138> andythenorth, shall I PR it?
22:33:10 <peter1138> Zoolook is energetic. Maybe not the best music to be listening to before bed.
22:35:34 <peter1138> Mmm, that gouda has a lovely flavour.
22:36:14 <andythenorth> peter1138: PR it!
22:37:17 <peter1138> Bed soon, rather than, say, 3am like last night.
22:37:51 <Eddi|zuHause> that never worked when i tried
22:38:16 <andythenorth> oh look in my diary
22:38:16 <peter1138> Eddi|zuHause, I'll get the kindle out for a change, instead of playing on the phone.
22:38:28 <andythenorth> 'risk register review'
22:38:39 <andythenorth> my glamorous life
22:38:55 <andythenorth> peter1138: I am reading Biggles :P
22:38:58 <andythenorth> for random reasons
22:43:21 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN opened pull request #7065: Change: Make ships stop and change direction slowly instead of instantly turning. https://git.io/fhWk0
22:44:34 <Eddi|zuHause> that immediately pops a bunch of thoughts in my mind
22:44:52 <DorpsGek_II> [OpenTTD/OpenTTD] andythenorth commented on pull request #7065: Change: Make ships stop and change direction slowly instead of instantly turning. https://git.io/fhWkK
22:45:34 <Eddi|zuHause> like, river ships making 45° turns more quickly and stuff
22:45:56 <LordAro> andythenorth: giv video
22:46:15 <peter1138> Eddi|zuHause, turn speed could be a property, currently it's fixed.
22:46:52 <peter1138> (And not entirely predictable, as it uses the vehicle tick counter, so may not be starting from zero)
22:47:07 <Eddi|zuHause> also unrelated: rivers need a flowing speed, so you can go faster downstream than upstream
22:47:22 <peter1138> They also need a direction, then.
22:47:44 <Eddi|zuHause> i meant a speed vector
22:48:24 <dwfreed> use the right physics term so we understand what you mean
22:49:06 <Eddi|zuHause> when you stop trying to make two words that mean the same thing into meaning different nuances
22:49:19 <Eddi|zuHause> that are entirely arbitrary
22:49:45 <dwfreed> physics has had these definitions for many, many years
22:50:04 <andythenorth> a velocity isn't a speed vector? :o
22:50:33 <dwfreed> I mean, it is, but a "speed vector" is a confusing way of saying it, when "velocity" already exists
22:51:31 <andythenorth> do rivers need a direction?
22:51:36 <andythenorth> can we give them some track bits
22:51:51 <Eddi|zuHause> dwfreed: how come "velocity vector" is then a word i find in my dictionary?
22:51:52 <andythenorth> oh we can do the direction as a byte?
22:52:18 <andythenorth> there are 8 directions?
22:52:35 <Eddi|zuHause> "velocity" in no way implies being vectorized any more than "speed" does...
22:52:55 <andythenorth> I was taught differently in physics eddi :o
22:53:09 <dwfreed> "Velocity is a physical vector quantity"
22:53:10 <Eddi|zuHause> just the english language has this habit of making up two words for the same thing
22:53:19 <Eddi|zuHause> like Beef and Cow
22:53:37 <Samu> im changing opf to be more intelligent at being bad
22:53:39 <Eddi|zuHause> and then pretending they don't mean the same thing
22:53:42 <andythenorth> except in a physics exam, I would usually specific 'velocity, as a vector'
22:53:59 <andythenorth> because velocity and speed are used interchangeably in plain speecvh
22:54:50 <andythenorth> so ship has upstream and downstream speeds?
22:54:51 <AKTheKnight> Beef vs cow was because of the normans
22:54:56 <AKTheKnight> Beef was what you ate
22:55:00 <AKTheKnight> Cow was the live animal
22:55:23 <peter1138> And that ... was a nice soft cheese.
22:55:32 <andythenorth> beef is often steer
22:55:54 <Samu> opf is using track to increase it's depth
22:55:55 <dwfreed> often times farmers refer to their cows as "beef" when that's what they intend to turn them into
22:56:02 <Samu> now i'm changing it to trackdir instead
22:56:10 <peter1138> Although I only had a small piece obviously, not the whole cheese.
22:56:24 <AKTheKnight> Sounds like you ate the whole thing to me
22:56:39 <Eddi|zuHause> dwfreed: in original french, "boef" means both the animal and the meat
22:56:43 <Samu> should go at greater depths now on horz/vert tracks
22:57:01 <Samu> if that results at it being smarter, is what's i'm gonna discover now
22:57:31 <Eddi|zuHause> in german it's slightly more complicated, because both the term "Kuh" and "Rind" exist
22:57:46 <peter1138> AKTheKnight, about 30 grams, so not a lot.
22:59:11 <peter1138> Ooh decent looking cheddar.
22:59:39 <peter1138> Gotta admit I tend to just go with the supermarket stuff for cheddar but maybe I should change.
22:59:58 <peter1138> Nothing really special about the premium brands in supermarkets.
23:00:00 <Eddi|zuHause> dwfreed: anyway, that wikipage is in no way a convincing argument that "speed" always means scalar and "velocity" always means vector...
23:00:13 <AKTheKnight> Well tbh I took this from my parents after Christmas. I've got the Waitrose version not the neals yard one
23:00:40 <peter1138> Mine is from Tesco :p
23:01:03 <AKTheKnight> Haha I usually get an asda shop, they're not known for their premium cheese :P
23:02:16 <peter1138> Hmm, the other cheeses from that link look nice too.
23:02:18 <andythenorth> Eddi|zuHause: in plain English, they're used interchangeably
23:02:39 <andythenorth> but if you did an A-level Physics exam paper, you would be expected to treat one as scalar, and one as vector
23:02:45 <AKTheKnight> Yeah I think their only shop is next to borough market, they do some fancy cheese
23:02:46 <Eddi|zuHause> andythenorth: and we're not using plain english here?
23:03:06 <andythenorth> it's on a par with the stupid BODMAS rules or whatever they're called
23:03:11 <andythenorth> for order of operations
23:03:25 <andythenorth> people who do it 'correctly' are adamant there is only one way
23:03:31 <Eddi|zuHause> in a scientific paper/exam i can make an arrow above the v symbol to make it clear i'm talking about the vector
23:03:33 <andythenorth> and then there's everyone else
23:03:45 <andythenorth> similar to the order of ops rules
23:03:53 <andythenorth> anyone doing engineering would put in the brackets
23:04:12 <andythenorth> and stop relying on people remembering dumbass acronyms
23:04:28 <Eddi|zuHause> people who think there is only one order of operator precedences haven't switched between programming languages enough
23:04:32 <peter1138> Right, night night.
23:04:48 <Eddi|zuHause> i don't understand cheese
23:04:59 <peter1138> Why do my passions have to be unhealthy?
23:05:07 <andythenorth> peter1138: don't expand that further pls :P
23:05:35 <andythenorth> I was made to feel stupid once for putting the brackets in
23:05:47 <dwfreed> "Please Excuse My Dear Aunt Sally"
23:05:55 <andythenorth> I was told it was redundant, and wasted space
23:06:16 <Eddi|zuHause> mnemonics for stupid rules are even worse than the stupid rules themselves
23:06:29 <andythenorth> except for SOHCAHTOA
23:06:53 <andythenorth> trig is the best maths
23:07:19 <Eddi|zuHause> you haven't done functional analysis then :p
23:08:36 <Eddi|zuHause> "let's take all these concept about vectors and dimensions, throw it all out, and redefine it in counterintuitive ways to make it work for infinitely many dimensions"
23:09:33 <andythenorth> I sat next to someone who did :P
23:09:43 <Eddi|zuHause> "what do you mean these spheres have all right angles on every surface point?"
23:09:58 <andythenorth> n-dimensional vectors in non-euclidean geometries
23:10:22 <Eddi|zuHause> n-dimensional is not infinite-dimensional :p
23:10:28 <Eddi|zuHause> you can make this even worse :p
23:10:40 <andythenorth> that raises the game
23:10:53 <andythenorth> finitely, but also infinitely
23:11:02 <andythenorth> it's a single step from finite to infinite
23:11:13 <Eddi|zuHause> not to mention that there are different levels of infinite
23:11:31 <andythenorth> that can only be discussed in Thursdays
23:11:46 <andythenorth> child #1 watches videos about it
23:12:14 <andythenorth> why did I do Hilbert spaces in philosophy?
23:12:38 <Eddi|zuHause> maths is the highest form of philosophy
23:13:32 <andythenorth> something about complex numbers and hilbert spaces
23:13:39 * andythenorth didn't really pay attention at the time
23:14:01 <Eddi|zuHause> you can also extend hilbert spaces to infinite dimensions
23:14:39 <andythenorth> something to do with decomposing the imaginary part
23:15:11 <andythenorth> my maths isn't as strong as it should be :P
23:15:47 <Eddi|zuHause> one of the "fun" bits about functional analysis: in "regular" maths you can make sums over infinitely many parts, as long as the "infinite" thing is countable. but in functional analysis you can generalize that even further, to sum over an uncountable infinite set, as long as only countably-many elements are non-zero
23:17:34 <Eddi|zuHause> andythenorth: complex numbers are a very easy special case of a hilbert space
23:29:06 *** wodencafe has joined #openttd
23:41:19 <Samu> i think i made opf too smart now lol
23:41:54 <Samu> it's fun to test things on it
23:43:59 * LordAro tempted to make a PR removing OPF
23:44:40 <Samu> no, i mean this tweak seems to have good results
23:44:50 <Samu> let me try the 5k ship map
23:47:25 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro updated pull request #7057: Fix: A few minor compile warnings under MinGW https://git.io/fhn6E
23:47:25 <Samu> i changed Track into Trackdir and adjusted accordingly, now it can increase it's search depth on horizontal and vertical, not just on diagonals
23:48:11 <Samu> the results are surprising me
23:48:26 <Samu> didn't expect anything of it, but it almost looks smart
23:49:26 <Samu> the way is was doing the searchs before were more fitting of a road vehicle
23:49:58 <LordAro> peter1138: #7065 needs rebasing to get CI results, btw
23:54:39 <Samu> uh oh, bad news, this "tweak" made it much more cpu insensive
23:55:02 <Samu> I used to get 90 ms avg, now it's 220 ms avg
23:56:12 <Samu> it goes through many asserts, i guess if i skip the asserts, it may be faster?
23:58:06 <dwfreed> asserts are generally removed in optimized builds
continue to next day ⏵
