Times are UTC Toggle Colours
00:07:55 *** Soni has quit IRC 00:50:53 *** Soni has joined #openttd 02:11:23 *** Wormnest has quit IRC 02:36:57 *** glx has quit IRC 02:50:33 *** D-HUND has joined #openttd 02:53:53 *** debdog has quit IRC 03:26:05 *** Smedles has quit IRC 03:29:05 *** Smedles has joined #openttd 04:01:46 *** HerzogDeXtEr1 has joined #openttd 04:08:08 *** HerzogDeXtEr has quit IRC 04:15:03 *** D-HUND is now known as debdog 06:02:35 *** sla_ro|master has joined #openttd 06:03:47 *** Tirili has quit IRC 06:28:20 *** Flygon has joined #openttd 07:20:29 *** supermop_work has quit IRC 08:35:36 *** lobster has quit IRC 08:41:42 *** lobster has joined #openttd 08:47:38 *** nielsm has quit IRC 09:35:17 *** gelignite has joined #openttd 09:51:59 *** Samu has joined #openttd 09:56:32 <Samu> hi 09:59:20 *** Samu has quit IRC 10:00:51 *** Samu has joined #openttd 10:23:30 *** sla_ro|master has quit IRC 10:38:23 <Samu> 2 images, can you spot the difference? https://imgur.com/a/UCNq42t 10:40:53 <Samu> the test was loading the same savegame and count how many pathfinder calls are done for the road vehicles till the start of next year 10:41:34 <Samu> savegame date was 13th nov 1945 10:43:58 <Samu> apparently it looks like there's no differences, but actually there are 10:48:01 <Samu> more testings tell: the clearing of legitimally valid path caches may result in more pathfinder calls. And even if the road network never changes, the pathfinder apparently does take into account vehicles blocking and can opt for an alternate path 10:48:18 <Samu> than the one it had cached 10:48:20 <Samu> before 10:49:00 <Samu> very interesting 10:49:22 <peter1138> What do you think the cache is for? 10:51:42 <Samu> so, getting another path when a vehicle is blocked by another is by design? 10:51:59 <TrueBrain> I am a bit scared how you navigate in traffic now :D 10:52:17 <peter1138> Sure, the pathfinders have always taken other traffic into account. 10:52:45 <LordAro> hmm 10:52:51 <LordAro> do i want to spend £500 on a chair? 10:53:00 <TrueBrain> yes 10:53:00 <TrueBrain> no 10:53:02 <TrueBrain> maybe 10:53:09 <peter1138> Do I want to spend £500 on a set of software synthesizers, or write my own? 10:54:12 <LordAro> TrueBrain: ta 10:54:19 <TrueBrain> always 10:55:30 <peter1138> The pathfinder cache is a tradeoff between being dynamic and 'immediately' chosing the best path, and performance. 10:57:07 <Samu> i thought that was a thing for trains 10:57:13 <Samu> not road vehicles, :( 10:59:31 <peter1138> That's one of the reasons the path cache is limited in the number of steps. 10:59:40 <peter1138> Can't remember if that's tiles or junctions. 11:06:43 <Samu> opntitle.dat savegame is a good example too, there are 3 different busses models 11:08:19 <Samu> the slower bus "blocks" the faster bus behind, and sometimes when on a junction, it decides to take another path, to avoid the slower bus 11:09:07 <Samu> interesting, i need to investigate the pathfinder itself, see where it does that block detection thing 11:16:20 <Samu> https://i.imgur.com/VHPtmYZ.png 11:16:46 <Samu> road vehicle 2 is the slow bus 11:16:53 <Samu> btw, drive is on left 11:17:06 <Samu> road vehicle 1 is behing road vehicle 2 11:17:21 <Samu> on 12.2 (or master), the fast bus turns left 11:17:46 <Samu> on my pr, since i made it not invalidate the cache, it continues following the slow bus 11:18:01 <Samu> it avoids a pathfinder call 11:18:12 <Samu> but choses a poor route 11:40:18 <peter1138> Okay, seems the PF only cares about curves, level crossings and road stop occuption. 11:41:02 <peter1138> Though there might be other tests than those explicit ones. 11:45:02 <TrueBrain> I still love about NewGRF specs, that the source often tells a better story :D 11:45:03 <TrueBrain> hihi 11:45:13 <peter1138> Or a misinterpretation of the story. 11:45:33 <TrueBrain> well, when it comes to "what bytes should I write" 11:45:39 <TrueBrain> it is a bit more clear than the specs :) 11:48:04 <TrueBrain> I think I can render Action0 for stations now ... now to build the rest to actually test that :P 12:06:49 *** sla_ro|master has joined #openttd 13:00:22 *** Flygon_ has joined #openttd 13:01:39 *** Flygon has quit IRC 13:03:38 *** tokai has joined #openttd 13:03:38 *** ChanServ sets mode: +v tokai 13:07:18 *** tokai|noir has quit IRC 13:21:25 <Samu> there's nothing in yapf regarding detection of another vehicle blocking 13:21:50 <Samu> then why does yapf come up with a different path? :( 13:23:11 <Samu> also the destination station isn't occupied 13:27:40 *** Smedles has quit IRC 13:27:46 *** Smedles has joined #openttd 13:33:03 <Samu> what's the max speed of road depots? unlimited? 13:49:32 *** glx has joined #openttd 13:49:32 *** ChanServ sets mode: +v glx 15:44:50 *** frosch123 has joined #openttd 16:29:05 *** Smedles has quit IRC 16:29:12 *** Smedles has joined #openttd 16:33:38 *** Smedles has quit IRC 16:33:43 *** Smedles has joined #openttd 16:48:07 <Samu> aha, i found some issue with yapf 16:48:15 <Samu> regarding costs 16:48:44 <Samu> the starting tile isn't being accounted for costs 16:49:11 <Samu> now i see why yapf was chosing another path 16:50:27 <Samu> the starting node needs a cost 16:53:53 *** Wormnest has joined #openttd 17:04:45 *** tokai|noir has joined #openttd 17:04:45 *** ChanServ sets mode: +v tokai|noir 17:08:14 *** tokai has quit IRC 17:08:44 <Samu> https://i.imgur.com/ucKnWfq.png the starting tile has a 0 cost 17:09:26 <glx> doesn't matter 17:09:54 <Samu> going left side, the total cost to reach station is 671 17:10:06 <Samu> going right side, the cost is 742 17:11:06 <Samu> if starting tile had a cost, it would be +171 for left side and +100 for right side 17:11:33 <Samu> they would tie at 842 17:14:14 <Samu> looks like this is what happens when the cache isn't purged, that portion costs is 842 17:14:43 <Samu> since it's a tie, it picked whichever came first, happened to be the right side one 17:17:33 <Samu> it has no detection of being blocked in the pathfinder calculations 17:17:44 <Samu> it was the result of costs being fishy 17:34:11 <DorpsGek> [OpenTTD/OpenTTD] Krakabraka opened issue #9916: [Bug]: Trains become lost and randomly reverse direction despite there being no signal https://github.com/OpenTTD/OpenTTD/issues/9916 17:49:58 <DorpsGek> [OpenTTD/OpenTTD] glx22 commented on issue #9916: [Bug]: Trains become lost and randomly reverse direction despite there being no signal https://github.com/OpenTTD/OpenTTD/issues/9916 18:23:03 *** nielsm has joined #openttd 18:25:57 *** Flygon__ has joined #openttd 18:31:13 *** Wolf01 has joined #openttd 18:33:37 *** Flygon_ has quit IRC 18:34:30 <peter1138> They're lost because there's no signal. Gotta wait for the GPS... 18:34:43 <LordAro> hehe 18:49:30 <DorpsGek> [OpenTTD/OpenTTD] DorpsGek pushed 1 commits to master https://github.com/OpenTTD/OpenTTD/commit/8cdf303e72ed133bbcc14898f73b1249b047af7f 18:49:31 <DorpsGek> - Update: Translations from eints (by translators) 18:50:26 <peter1138> Hmm, why is my 8 channel sound card not enough suddenly? 18:50:45 <frosch123> maybe you grew more ears? 18:55:25 <TrueBrain> lol 20:12:21 *** WormnestAndroid has quit IRC 20:17:31 *** Flygon__ has quit IRC 20:36:23 <wallabra> Wait for iiiiit... https://i.imgur.com/PpoP1ML.png 20:46:23 *** nielsm has quit IRC 21:15:11 *** gelignite has quit IRC 21:48:23 *** Wolf01 has quit IRC 21:48:52 *** frosch123 has quit IRC 21:59:44 *** Samu has quit IRC 22:03:12 *** sla_ro|master has quit IRC 23:24:36 *** wallabra has quit IRC 23:44:54 *** WormnestAndroid has joined #openttd 23:45:10 *** Alkel_U3 has quit IRC 23:47:52 *** Alkel_U3 has joined #openttd 23:50:14 *** wallabra has joined #openttd