IRC logs for #openttd on OFTC at 2024-10-01
⏴ go to previous day
00:17:40 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
00:35:23 <wensimehrp> belajalilija: general suggestions? idk what JF is doing there
00:36:37 <wensimehrp> maybe he can fetch the member list of this discord server and dump it into a csv file and add it to the repo
00:37:46 <wensimehrp> 😎 maybe it's just because you're cool
01:03:38 *** Wormnest has joined #openttd
02:11:33 *** gnu_jj_ has joined #openttd
02:14:44 *** gnu_jj has quit IRC (Ping timeout: 480 seconds)
02:19:18 *** Wormnest has quit IRC (Quit: Leaving)
02:25:44 *** debdog has quit IRC (Ping timeout: 480 seconds)
05:02:14 *** keikoz has quit IRC (Ping timeout: 480 seconds)
05:09:23 *** keoz has quit IRC (Ping timeout: 480 seconds)
07:59:31 <wensimehrp> certainly not a good time for lunch
08:04:22 <kuhnovic> Somewhere in the world people are having lunch right now
08:25:50 *** SigHunter has joined #openttd
08:50:23 <xarick> if (path_cache.empty()) {
08:50:23 <xarick> if (!empty_path && node->GetTrackdir() != INVALID_TRACKDIR) return node->GetTrackdir();
08:50:23 <xarick> /* A empty path means we are already at the destination. The pathfinder shouldn't have been called at all.
08:50:23 <xarick> * Return a random reachable trackdir to hopefully nudge the ship out of this strange situation. */
08:50:24 <xarick> return CreateRandomPath(v, path_cache, 1);
09:31:59 <peter1138> "My phone has gone into DOS"
09:32:25 <peter1138> 1) your phone is probably bricked, 2) just because it's text on a black background doesn't mean it's DOS...
09:46:29 *** nielsm has quit IRC (Ping timeout: 480 seconds)
10:06:56 <xarick> i need track2 to match track in this specific case
10:07:53 <xarick> the tile that comes after trackdir_upper_e is trackdir_lower_e, glx was right
10:08:16 <peter1138> "What spacing does TTD use?"
10:32:24 <belajalilija> johnfranklin: i did?
11:02:52 <peter1138> vcpkg updated too, just needs a release I guess.
11:03:22 <peter1138> Microsoft is evil -> entire build chain depends on it. Ah well 😉
11:51:59 <kuhnovic> Do we always use the latest vcpkg?
11:54:15 <_glx_> We use whatever is bootstrapped
12:01:27 <kuhnovic> We could choose to clone a specific release. But then we'll forget to update 😛
12:01:48 <truebrain> Glad you ask and answer it yourself 😉
12:03:38 <kuhnovic> I'm just narrating my own thoughts here
12:04:00 <kuhnovic> It's what people do 😉
12:08:54 <truebrain> kuhnovic: Please don't make it a trend. One is plenty. We really don't need more. Please. No. Nooooo 😛
12:11:58 *** reldred has joined #openttd
12:11:58 <reldred> I try and solve that problem by not thinking
12:12:32 <andythenorth> I have paid a bot
12:12:36 <andythenorth> it listens better
12:12:45 <andythenorth> although it types WAY too much back
12:52:57 <andythenorth> internet -> pastes stack overflow commands for 15 years without understanding
12:53:06 <andythenorth> LLMs -> trained on internet
12:58:31 <_glx_> restarted some failed CI runs for a test, I can confirm vcpkg is working
13:00:39 <peter1138> Hmm, so it doesn't track (github-tagged) releases then.
13:02:46 <_glx_> they updated in vcpkg repo 🙂
13:03:15 <_glx_> and we always clone master
13:10:44 <LordAro> peter1138: screaming noises
13:17:16 <LordAro> the log is really something
13:17:40 <truebrain> peter1138: I just suspect someone fucked up their laptop, and used ChatGPT to write a fake story about how an LLM caused it; I just refuse to believe that actual story 😛
13:18:37 <LordAro> truebrain: i did wonder about apparently having homebrew on a ubuntu install
13:18:44 <LordAro> i suppose it's possible
13:20:55 <truebrain> We used to say: the dog ate my homework
13:21:03 <truebrain> Now it is: the LLM agent deleted my homework!
13:29:11 <peter1138> Quick before they break it again 😉
13:42:51 <xarick> In ChooseShipTrack, I need a way to distinguish ship entering new tile vs ship not entering new tile
13:44:01 <xarick> ` static Trackdir ChooseShipTrack(const Ship *v, TileIndex &tile, TrackdirBits forward_dirs, TrackdirBits reverse_dirs, const std::span<TileIndex> dest_tiles, int max_penalty,
13:44:01 <xarick> bool &path_found, ShipPathCache &path_cache, Trackdir &best_origin_dir, bool simulate_depot_order = false)`
13:49:46 <xarick> simulate_depot_order is temporary, for debugging
13:49:58 <xarick> will be removed in the finalized version
13:51:22 *** HerzogDeXtEr has joined #openttd
14:20:39 <andythenorth> wonder if getting rid of these will make git st faster? 😛
14:21:16 <andythenorth> they're junk pngs, generated by accident, my script has found them all and moved them to tmp
14:28:54 <LordAro> depends if they're suitably ignored
14:30:08 <andythenorth> I'm pruning them from history
14:32:17 <LordAro> ah, that sort of junk
15:25:20 <xarick> gettinh unhandled exception at thread ottd::game or something
15:30:03 <_glx_> and first part might be an accidental ctrl-clic
15:31:37 <_glx_> xarick: generic catch all, good luck to trace it 🙂
15:32:31 <xarick> seems to happen after failing to write savegame on crash?
15:32:46 <xarick> i'll post screenshots again when it happens
15:33:33 <_glx_> this unhandled exception stuff is almost impossible to trace without knowing the actions leading to it
16:04:31 <xarick> experimented removing SaveRandomSeeds
16:04:59 <xarick> and BinaryHeap without tie-breaker
16:05:07 <xarick> one of them is at fault
16:07:28 <xarick> I feel I'm reaching somewhere
16:09:28 <xarick> oh, i also removed the may_reverse code
16:11:38 <xarick> now let's see if this can live without SaveRandomSeeds and may_reverse
16:13:58 <_jgr_> _glx_: Arguably it'd be better to not have the try/catch at all
16:14:32 <_jgr_> That way you ought to get a proper stack trace from the crash logger
16:16:15 *** gelignite has joined #openttd
16:27:40 <xarick> game loop total has a spike that should belong to ship ticks
16:31:45 *** Smedles has joined #openttd
16:32:25 <xarick> Who's metering CheckIfShipNeedsService?
16:33:46 *** guru3 has quit IRC (Read error: Connection reset by peer)
16:46:24 <xarick> looks like feeding ~500 depots for each pathfinder call which is called each day per ship that needs servicing is a bad idea
16:46:58 <xarick> automatic service ruining my day 🙂
16:48:41 <xarick> but at least I got it into a state it's no longer asserting
16:55:00 <xarick> I need to approach the hlpf in a different manner...
16:58:02 <kuhnovic> Who would have though 😆
16:59:15 <xarick> I'm reviewing your method again
17:00:34 <xarick> I'm particularly interested in Step 1
17:07:23 <xarick> let me think... getting 500 depot_tiles isn't exactly expensive, the actual pathfinding with 500 origins isn't expensive, it's the act of adding up to 500 water patches that is expensive 😦
17:08:15 <xarick> need to optimize this step, and your step 1 seems to be the way
17:11:14 <xarick> how to quickly check each of the 500 depot_tiles whether they're within this "max_penalty"
17:11:41 <xarick> and only add origins of those that are
17:11:52 <xarick> so that we won't have 500 origins
17:22:45 <xarick> `if (find_closest_depot && high_level_max_penalty != 0 && DistanceMax(start_tile, tile) > high_level_max_penalty / YAPF_TILE_LENGTH) continue;`
18:00:42 <xarick> i no longer feed 500 depot_tiles
18:02:07 <xarick> max_distance is max_penalty / YAPF_TILE_LENGTH
18:02:47 <xarick> the list is forwarded to hlpf and llpf
18:47:43 <andythenorth> peter1138: but you did?
18:48:04 <peter1138> I was paraphrasing the poster.
18:50:25 <xarick> I failed! I'm disappointed
19:04:17 <andythenorth> lol this is so slow to do
19:04:24 <andythenorth> git is achingly slow
19:04:37 <andythenorth> git st takes 7 seconds 😛
19:04:55 <andythenorth> traversing paths and selective git rm is super slow
19:07:27 *** Wormnest has joined #openttd
19:15:18 <ahyangyi> Is that your `.git` directory?
19:15:21 <xarick> 1.9 GB = 3.57GB? is it compressing?
19:15:36 <andythenorth> running a script with git rm
19:15:59 <ahyangyi> That's a lot of files
19:16:49 <_glx_> xarick: cluster size, smaller files use lot of real space
19:16:52 <andythenorth> "mistakes were mad"
19:23:26 *** jijiji has quit IRC (Remote host closed the connection)
19:31:37 <andythenorth> hmm `git st` is getting faster now
19:32:21 *** gelignite has quit IRC (Quit: Stay safe!)
19:38:46 <_glx_> you had ton of untracked files ?
19:39:33 <_glx_> or too many tracked ones
19:43:14 *** tokai|noir has joined #openttd
19:43:14 *** ChanServ sets mode: +v tokai|noir
19:43:35 <andythenorth> way too many tracked
19:43:57 <andythenorth> my docs image generator was generating 10k unnecessary images per Horse release
19:44:20 <andythenorth> funny thing is, PIL was quite fast to do that, so I didn't notice during the compile 😛
19:45:38 *** Etua has quit IRC (Quit: Etua)
19:45:53 *** Etua has quit IRC (Remote host closed the connection)
19:47:34 <xarick> xarick: problem solved! automatic service can't search backwards. Now the ships happily don't service :p
19:49:45 <andythenorth> `git config feature.manyFiles true` might be worth a try
19:49:58 *** tokai has quit IRC (Ping timeout: 480 seconds)
19:50:05 <andythenorth> I have been using `git config core.fsmonitor true` so far
20:09:38 <xarick> checking for asserts + bad behaviour
20:11:45 <xarick> i dont understand why there's a ? in one of them... they're the same executable
20:13:34 <andythenorth> hmm not sure github loves my commit with 30k files in it 😛
20:17:32 <_pruple> there is no idea, no matter how inane, that cannot get zorged D:
20:20:52 <_pruple> a setting that prevents you changing any other settings until you change that setting, what a concept mark. 😄
20:24:31 <andythenorth> that would need a setting also
20:24:36 <andythenorth> can I see the setting setting
21:09:29 <andythenorth> have deleted 110k files so far 😛
21:09:34 <andythenorth> that's about 20%
21:24:41 <andythenorth> much faster now it's 370k files
21:24:47 <johnfranklin> peter1138: It did some sudo rm -rf *?
21:25:39 <johnfranklin> _pruple: Such “don’t play with your phone” apps on a phone
21:27:02 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
21:31:38 *** nielsm has quit IRC (Ping timeout: 480 seconds)
21:33:09 *** keikoz has quit IRC (Ping timeout: 480 seconds)
21:36:54 <xarick> xarick: lol, i had an error in this code
21:37:14 <xarick> nobody else noticed either
21:38:25 <xarick> `if (max_distance == 0 || DistanceMax(depot_tile, tile) <= max_distance) depot_tiles.push_back(depot_tile);` this is what it should have been
21:40:35 *** Wolf01 is now known as Guest5141
21:46:04 *** Guest5141 has quit IRC (Ping timeout: 480 seconds)
21:46:32 <xarick> alright, testing asserts overnight
21:51:38 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
22:41:24 <peter1138> > nobody else noticed
22:41:32 <peter1138> > nobody else looked
23:36:18 <_pruple> is it time to set "scoring end year" to "never" by default?
continue to next day ⏵
