IRC logs for #openttd on OFTC at 2025-03-16
⏴ go to previous day
00:46:29 <truebrain> LordAro: totally not unexpected, the reply I got from Steam:
00:46:31 <truebrain> ```We have chosen to limit access to some community features as a means of protecting users from those who abuse Steam merely for purposes such as spamming and phishing.
00:46:31 <truebrain> To access all features of Steam, you will need to spend at least $5.00 USD within the Steam store.```
00:46:38 <truebrain> I doubt they even read anything I wrote
00:46:45 <truebrain> and just hit the "auto reply with template" button
00:49:27 <truebrain> "It works via the browser but not via steamcmd" - "owh, your account has limit access"
00:49:32 <truebrain> In what world does that make sense?
00:52:18 <truebrain> Ah, also found a support portal for the Partner part ... let's ask the same question there!
00:54:25 <truebrain> (and no, I am not going to pay 5 USD in the "hope it solves a problem" 😛 )
01:04:24 <jfkuayue> I thought they were powered by AI
01:34:15 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
01:34:17 *** WormnestAndroid has joined #openttd
01:55:46 *** Tirili has quit IRC (Remote host closed the connection)
02:11:12 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
02:12:47 *** gelignite has quit IRC (Quit: Stay safe!)
02:13:02 *** WormnestAndroid has joined #openttd
03:05:39 *** gelignite has joined #openttd
03:05:54 *** gelignite has quit IRC (Remote host closed the connection)
03:09:31 *** D-HUND has quit IRC (Ping timeout: 480 seconds)
03:17:22 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
03:17:24 *** WormnestAndroid has joined #openttd
03:17:27 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
03:17:42 *** WormnestAndroid has joined #openttd
03:32:36 *** geizeskrank has quit IRC (Ping timeout: 480 seconds)
03:36:09 *** geizeskrank has joined #openttd
03:48:51 *** Tirili has quit IRC (Remote host closed the connection)
04:07:01 *** debdog has quit IRC (Ping timeout: 480 seconds)
05:20:49 *** Extrems has joined #openttd
07:16:59 <peter1138> It is far too cold.
07:17:30 <reldred> we just had a cool change roll through today, it was 40C yesterday
07:20:02 <reldred> my knees would ache in that weather
07:47:22 *** Tirili has quit IRC (Remote host closed the connection)
08:24:46 *** nielsm has quit IRC (Ping timeout: 480 seconds)
09:04:28 <_zephyris> truebrain: Isn't there a one off fee to publish anyway?
09:05:12 <truebrain> There is. What makes me very sus of this answer
09:05:42 <truebrain> But who knows, they might now restrict business accounts to their steamcmd because it hasn't any transaction ......
09:06:38 <_zephyris> Mmm, plausible and annoying
09:08:43 <michi_cc> So we are supposed to publish a 1 cent golden toilet DLC now? 🤣
09:11:35 <truebrain> funny enough, my own OpenTTD account, I can login via SteamCMD just fine. And has the same restrictions
09:12:04 <truebrain> The only difference is that it uses Steam Mobile to do 2FA, and the upload account uses an OTP
09:12:52 <truebrain> so I am somewhat tempted to just create a new Steam account 😛 😛
09:14:33 <truebrain> mostly, it is still very silly a platform like Steam doesn't have API keys to allow uploading new versions of your game. That is still mind blowing to me.
09:19:24 <peter1138> Do the regression tests test cargo monitors?
09:21:40 <peter1138> Or even... what GS use it.
09:28:50 <_zephyris> Custom industry fields and coffee plantations...
09:35:22 <peter1138> /* Skip the cargo type section, we only care about the default group */
09:35:39 <peter1138> Industries don't have per-cargo spritegroups.
09:37:17 *** wallabra has quit IRC (Read error: Connection reset by peer)
09:37:36 <peter1138> What if we appropriate an entry for field graphics.
09:37:57 <peter1138> But uh, is it just a ground sprite, or is it tile layout?
09:38:27 *** wallabra has joined #openttd
09:38:47 <_zephyris> Preferably a tile layout
09:39:07 <_zephyris> Plants, trees, greenhouses
09:39:11 <peter1138> There are 19 different slopes, so "a tile layout" isn't exactly right.
09:39:53 <_zephyris> Could restrict it to a single ground sprite, single above ground sprite
09:40:06 *** kuka_lie has joined #openttd
09:40:21 <_zephyris> And not go full sprite layout, which might tank performance if abused
09:42:24 <peter1138> Hmm, searching an unordered std::vector vs std::map... less than idea :/
09:44:51 <_zephyris> Fences too, set of a multiple of 6
09:45:48 <_zephyris> Field 'animation' too, actually a set of 9(?)x19 ground tiles.
09:51:50 <peter1138> For the "animation" state, I would let the NewGRF switch on that.Se
09:52:11 <peter1138> Seems like there's only 8 stages, even though 4 bits are used (which would allow 16 stages)
09:53:07 <peter1138> Or hay stacks are handled specially.
09:53:45 *** kuka_lie has quit IRC (Quit: Lost terminal)
09:53:48 <peter1138> Ah no, 9 stages, indeed.
09:54:21 <peter1138> Anyway, 16 could be possible.
09:55:42 <peter1138> Someone will want company-coloured fields, even though industries are not company-owned...
09:57:09 <andythenorth> red ones, blue ones
09:57:54 <frosch123> I kind of liked the idea, that industries could spawn objects, which look like fields, surface mining, waste dumps, ...
09:58:22 *** kuka_lie has joined #openttd
09:59:07 <peter1138> Handling slopes with objects seems to be quite a pain.
10:00:12 <_zephyris> Only if you don't want foundations...
10:00:15 <frosch123> Fields are pure visuals with no function. So reinventing a new drawing code for them is like station non-track tiles
10:02:52 <_zephyris> Kinda... But it does give a visually consistent way of making it more expensive to build in big tracts of the landscape
10:02:54 <peter1138> Hmm, could tile layouts gain a "draw the correct slope" flag...
10:03:46 <peter1138> Because authors seem to have difficulty with that. Not sure if it's the spec or the documentation, mind you.
10:04:44 <_zephyris> It just takes a single switch, but is a moderately complex one
10:08:28 *** gelignite has joined #openttd
10:09:41 <xarick> @rubidium, there are some XXX comments, they're some kind of todo
10:11:15 <peter1138> I don't know any more. Okay, this is not my side-quest, other people already have other better ideas.
10:12:02 <frosch123> Since when is drawing the sloped sprites the problem? I only know it gets complicated if you want separate buildings with foundations on different elevations
10:13:05 <_zephyris> It's not a problem, but for the (no offence) grf authors who churn out object sets it seems too complex
10:16:23 <peter1138[d]> Not sure what this was about any more, but someone was having 'fun' with slopes...
10:16:57 <peter1138> Possibly the combo of slope & water checks rather than just slopes.
10:17:50 <peter1138> _zephyris, quantity not quality.
10:22:43 <wensimehrp> peter1138[d]: I wasn't
10:23:06 <wensimehrp> (yes I do see the quotation marks)
10:25:17 <_zephyris> Translating slopes to coasts is always a bit fiddly
10:30:06 <peter1138> Okay so plain slopes are easy with tile layouts, but objects on coasts are not?
10:31:03 <_zephyris> Mildly harder, and only if you want to somehow retain/represent water on the watery corners
10:33:48 <_zephyris> Hmm, IIRC, it's differentiating between sea level and wet tiles
10:34:01 <_zephyris> Now I'm thinking I can't quite remember
10:38:39 <xarick> shall I stop the test? it doesn't seem to get much better than this. or shall I wait for the profits to condense a bit more?
10:40:06 <xarick> i haven't restarted windows since 10th of march due to this test
10:40:37 <xarick> and windows wants to update stuff...
10:41:51 <LordAro> i'm pretty sure only you can answer that question
10:44:18 <xarick> im kinda sad AAAHogEx doesn't really aim for 5000
10:45:02 <xarick> seems to be doing some profit managements at the limit
10:45:22 *** kuka_lie has quit IRC (Read error: Connection reset by peer)
10:48:38 <cu-kai> got a picture of the map?
10:50:46 <cu-kai> what CPU are you using?
10:52:32 <cu-kai> single core speed around the same as my i9
10:52:39 <cu-kai> too bad openttd can't be multithreaded
10:56:32 <xarick> 4k map is enough for 75k trains
10:56:45 <cu-kai> only if you want to play at 1fps
10:56:46 <xarick> ppl still want bigger maps yet
10:57:06 <cu-kai> the largest i generated for a private multiplayer server was 1k
10:58:48 <kuhnovic> A 4K is already gigantic, I just don't understand that people want to go even bigger.
10:59:11 <cu-kai> lol, did i read here that people wanted 16k maps? :D
10:59:25 <kuhnovic> But as with any limit, the moment you come up with one there's always somebody that wants to see it doubled.
10:59:25 <cu-kai> the comment was some days ago, or longer
10:59:40 <truebrain> Ugh. Looking further into this Steam bla, and was looking over our actions ... CodeQL is totally trashing our attempt to cache stuff. You can only have 10GB of stuff in your Cache, and each CodeQL run creates 1GB in Cache. So yeah ... our cache is rather empty 😛
10:59:48 <kuhnovic> cu-kai: JGRPP already supports it
11:00:47 <cu-kai> huh, this is an interesting project
11:00:56 <truebrain> (mostly, our vcpkg is no longer in cache, it seems)
11:01:47 <truebrain> `Restored 12 package(s) from GitHub Actions Cache`
11:02:06 <truebrain> so the cache overview tells me: you don't have vcpkg in your cache
11:02:12 <truebrain> but the run itself says: sure, I got it from the cache!
11:04:28 <cu-kai> > Extra large maps. Maximum map size is now 256M tiles, ranging from 16k x 16k to 256 x 1M.
11:05:11 <xarick> does JGR increase max num of vehicles, orders, stations, etc...
11:06:07 <xarick> I already struggle with station and order pool limits with my AI
11:08:08 <xarick> i am forced to use shared orders to avoid reaching orderlist limits
11:08:57 <xarick> limit train sizes to about 5 or 6 to avoid vehicleid limits
11:09:25 <xarick> so, 16k maps without these other values scaling as well feels pointless
11:11:15 <xarick> anyway, 1 fps per second is ... meh
11:12:22 <xarick> it's 2 tall tylers fault :=)
11:12:40 <truebrain> Right, that PR should fix that issue. Not the problem I was looking for, but the problem I found. Side-quests!
11:13:02 <xarick> tick length was reduced, so everything needs to compute more often
11:13:55 <xarick> it's totally not my fault that I'm running 75k trains, 🙂
11:14:59 <peter1138> Not everyone is trying to max out the limits.
11:15:53 <_jgr_> xarick: The per-company vehicle limits and per-vehicle order limits are increased. There isn't any need to increase the actual pool sizes.
11:17:00 <_jgr_> xarick: The point of larger maps is not to fill every available space with vehicles, if anything it's the opposite, to make things slower and more spread-out
11:18:51 *** kuka_lie has joined #openttd
11:21:55 <andythenorth> wonder what the ultimate map size is
11:21:59 <andythenorth> 64x64 is too small
11:22:14 <andythenorth> although it would make it plausible to use GS to control landscape per tile
11:25:25 <xarick> let me take a look at what I've done
11:30:19 <xarick> > StationID: 64k -> 0xFF000, 64k per company
11:30:19 <xarick> > DepotID: 64k -> 0xFF000
11:30:19 <xarick> > GroupID: 64k -> 0xFF000, 64k per company
11:30:19 <xarick> > OrderListID: 64k -> 0xFF000
11:30:19 <xarick> > RoadStopID: 64k -> 0xFF000
11:30:29 <xarick> apparently I didn't increase VehicleID limits 😛
11:33:18 <xarick> if the average aircraft takes 2 vehicle ids, average ship takes 1, average roadveh takes 1, then that takes 300k vehicleids.
11:33:36 <xarick> about 700k remaining for trains
11:34:48 <xarick> about ~9 vehicles per train, that's train sized 4.5 tiles
11:36:10 <xarick> I have the ultimate dream of 75k of every vehicle type
12:02:31 <jfkuayue> 64x1048576 would be interesting to play
12:02:48 <peter1138> I'd like to use something like PoolID but for non-pool items. What to do...
12:03:12 <peter1138> Perhaps StrongType is more appropriate.
12:04:33 <ahyangyi> jfkuayue: 🤔 Perhaps I should make a 16:1 Curveland 5?
12:05:14 <peter1138> jfkuayue, I bet it would not be interesting to play.
12:07:25 <xarick> dear.. someone, I can't build clang-release with "configurationType": "Release",
12:07:40 <xarick> only with "configurationType": "RelWithDebInfo",
12:08:23 <xarick> where is this manifest file
12:11:04 <_glx_> Release and RelWithDebInfo should be very similar
12:14:59 <xarick> gonna try delete the entire vcpkg
12:19:47 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
12:21:21 <xarick> deleted the out folder, let's see now
12:22:47 <xarick> nope, doesn't work, still same error
12:27:00 <xarick> windows events tells me this
12:31:07 <xarick> when I switch to RelWithDebInfo, it runs
12:31:22 <xarick> seems to be missing the reqestedExecutionLevel
12:31:34 <xarick> according to a bing search
13:02:14 <xarick> how do i check whether the manifest is being... uh... looked upon?
13:03:10 <xarick> i see some .ps1 and .bat scripts here
13:14:25 <kuhnovic> jfkuayue: As a dutchman I approve
13:27:18 <_glx_> oh weird it fails for me too
13:27:46 <digitalfox> I was curious and looking at all the open PR…
13:27:46 <digitalfox> I’m no Dev and who am I to judge…
13:27:46 <digitalfox> But Xarick, you have 28 open PR, are you sure they are all still relevant?
13:32:37 <frosch123> Have fun reviewing that one :p
13:34:14 <Rubidium> frosch123: done... "needs work" :D
13:35:12 <ahyangyi> yeah, the CI didn't pass 😦
13:45:58 <_glx_> extracted from "broken" openttd.exe
13:47:49 *** WormnestAndroid has quit IRC (Remote host closed the connection)
13:47:50 *** WormnestAndroid has joined #openttd
13:52:16 <_glx_> RelWithDebInfo is "correct"
13:52:59 <xarick> where is this located?
13:53:41 <_glx_> needs to be extracted using mt.exe
13:54:40 <xarick> which of my 28 PRs are irrelevant
13:54:45 <peter1138> Hmm, need to revisit this design. Or simply not use this design.
13:54:45 <_glx_> anyway our manifest is fine
13:55:10 <peter1138> Storing widget state inside a widget is fine until you refresh the widgets and your state is lost. Oops.
13:56:01 <truebrain> oooeeehhhh, are we now using clang-tidy? 😄
13:57:36 <frosch123> Except compilers disagree on the result :)
13:58:18 <truebrain> the question is: is clang disagreeing?
13:58:21 <truebrain> as that would be funny 😛
13:58:29 <_glx_> I remember when we had a fight between gcc and clang
13:58:32 <frosch123> truebrain: I wonder who will side-quest the remaining 100+ tidy rules
13:59:05 <truebrain> if we can add clang-format to it too
13:59:14 <truebrain> we can just kill every patch out there, and be done with it 🙂
13:59:21 <truebrain> add tidy and format to the CI, and they never reappear!
14:05:03 <xarick> wow rubidium touched order flags...
14:07:40 <xarick> are you going to do something about unbunch for scripts?
14:08:25 <xarick> i'd use it for my ships
14:09:00 <xarick> cus they really bunch up on top of each other no matter what I try
14:35:12 <xarick> trying to understad HasAtMostOneBit
14:40:30 <xarick> ```case OT_GOTO_DEPOT:
14:40:30 <xarick> return (order_flags & ~(OF_NON_STOP_FLAGS | OF_DEPOT_FLAGS)) == 0 &&
14:40:30 <xarick> ((order_flags & OF_SERVICE_IF_NEEDED) == 0 || (order_flags & OF_STOP_IN_DEPOT) == 0) &&
14:40:30 <xarick> ((order_flags & OF_SERVICE_IF_NEEDED) == 0 || (order_flags & OF_UNBUNCH_IN_DEPOT) == 0) &&
14:40:30 <xarick> ((order_flags & OF_STOP_IN_DEPOT) == 0 || (order_flags & OF_UNBUNCH_IN_DEPOT) == 0);```
14:40:31 <xarick> How do I use `HasAtMostOneBit` here?
14:44:40 <xarick> `HasAtMostOneBit(order_flags & (OF_SERVICE_IF_NEEDED | OF_STOP_IN_DEPOT | OF_UNBUNCH_IN_DEPOT)` will be correct?
15:05:26 <xarick> oh HasAtMostOneBit also returns true for 0 bits
15:07:01 *** gelignite has quit IRC (Quit: Stay safe!)
15:07:22 <_glx_> no bits is at most one 🙂
15:13:25 <xarick> i understand how it works now
15:22:08 <xarick> 2cc trains in nml 3.1 made my AI crash
15:23:32 <xarick> railtype 255 is very suspicious
15:23:54 <xarick> local railtype = AIEngine.GetRailType(engine);
15:24:45 <xarick> assert(AIEngine.GetVehicleType(engine) == AIVehicle.VT_RAIL); this is where it fails
15:26:22 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
15:26:23 *** Wormnest has joined #openttd
15:26:24 *** WormnestAndroid has joined #openttd
15:26:38 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
15:26:41 *** WormnestAndroid has joined #openttd
15:26:42 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
15:26:43 *** WormnestAndroid has joined #openttd
15:29:06 <xarick> why do these newgrfs always make billions of trains
15:31:40 <truebrain> Rubidium: close 24, get 22 new ones? 😄
15:33:38 <Rubidium> mostly the new version of fmt it seems, but also... some of the issues show many instances so if you solve one instance, you solved one but the remaining instances seem to become new issues
15:33:56 <Rubidium> in any case... going into the right direction
15:35:06 <truebrain> Always forward! \o/
15:39:09 <Rubidium> and one annoying thing about fmt... we only use it in 'char` mode, so code that's there for `char16_t`/`char32_t`/`wchar` will trigger warnings about dead code or checks that'll always be true or false. So probably a lot of them can be dismissed
15:39:59 <xarick> problem must have been an engine became obsolete, removed from the list
15:45:56 <truebrain> can't you just ignore 3rdparty completely or something?
15:47:12 <Rubidium> yes, you can... but I rather dismiss a few false positives than miss issues
15:48:34 <kuhnovic> This one might kick up some dust 😛
15:49:01 <truebrain> Rubidium: Fair enough!
15:50:55 <truebrain> kuhnovic: Your PR is very unclear if this left behavior is on by default or behind the right click close setting
16:01:01 <kuhnovic> Clarified it a bit 🙂
16:01:41 <Rubidium> truebrain: 16 dismissed, the other 8 are in the std::move PR.
16:02:34 <Rubidium> only coverity usually doesn't seem to update the statistics after dismissing, only after uploading a new build
16:09:11 *** geizeskrank has quit IRC (Ping timeout: 480 seconds)
16:11:05 <truebrain> sorry kuhnovic ; generally I love your ideas. This really is not one of them. It really really really isn't 😄
16:11:22 <truebrain> Also: please play around with radical PRs first 😛 I broke it very easily 😄
16:11:37 *** geizeskrank has joined #openttd
16:12:27 <kuhnovic> A screaming toddler with a diaper full of number two came running in, I was a little distracted 😛
16:13:01 <truebrain> Owh, is that the excuse to not test PRs these days? 😛 😛
16:13:15 <truebrain> Anyway, I still think you are awesome. Just not this PR 😄
16:13:25 <kuhnovic> Nope, it's just my lame excuse for being a bit absent minded 😛
16:13:45 <kuhnovic> So why don't you like it, apart from it aparently not working as intented
16:13:53 <kuhnovic> (need to go now, I will look at it later)
16:13:57 <truebrain> in my book, it is just hostile UX
16:14:06 <truebrain> most errors are created by left click
16:14:12 <truebrain> so if you click twice, you create the error and close it
16:14:19 <truebrain> that will be very frustrating very quick
16:14:29 <truebrain> so I imagine your actual issue is not with the error window, but this specific usecase
16:14:35 <truebrain> as in that specific usecase, the error window is very annoying
16:15:27 <kuhnovic> Well I wasn't wrong about the pitchforks 😄 no hard feelings man, I appreciate you being honest. I'll get back to this later.
16:15:53 <truebrain> If you want to go this route, what you might do, is put a grace period in there, where the error doesn't close
16:15:56 <truebrain> so like 1 sec of not closing
16:16:02 <truebrain> but I am not sure that improves the situation
16:16:30 <truebrain> I also played with the idea of putting the error window just above the status bar, full width
16:16:35 <peter1138> I'd rather just promote right-click to close .
16:16:39 <truebrain> the height of the status bar
16:16:52 <talltyler> I wonder if a better resolution is showing fewer unnecessary errors in the first place.
16:17:22 <truebrain> I really don't like the error window jumps around like a mad man
16:17:25 <truebrain> that gets me so frustrated
16:17:35 <peter1138> toast-style popups?
16:18:14 <peter1138> Hmm, this drop-down list hack is... way to hacky :S
16:18:23 <peter1138> And not a Hackykid in sight.
16:18:25 <truebrain> So I do think there is reason to actually look into this stuff kuhnovic . As in general I do agree with you, especially with that specific usecase, that our error reporting is .. "meh" at best 🙂
16:18:46 <truebrain> Just please no conflicting UX patterns 😄 😄 😄
16:19:15 <_glx_> oh it's just another UX on top of all the existing ones
16:21:23 <_glx_> but yeah, not fan of left click to close
16:22:26 <truebrain> Now I want to open my box of crazy ideas how to deal with errors again 😛
16:22:43 <truebrain> I also toyed with making window UI elements red when there was an error; but that was horrible 😛
16:22:57 <truebrain> maybe combining it with peters idea of popups ... 😄
16:33:20 <xarick> locks make my hovercrafts slow
16:37:42 <xarick> China Discovers 60,000-Year Supply of Thorium
16:44:32 <_zephyris> peter1138: Wait, you can right click to close errors?
16:45:05 <_zephyris> (can't check on a computer, have my own toddler issues...)
16:45:10 <peter1138> You can right click to close (almost) any window, if the appropriate setting is set.
16:45:43 <peter1138> Exception is windows with viewports where right click scrolls.
16:45:58 <peter1138> (And even then that's only on the viewport itself.)
16:51:35 <truebrain> _zephyris: No matter your settings , right click always closes errors
16:51:55 <truebrain> I never managed to get used to it, it is weird
16:53:13 <_zephyris> Similar to blender IIRC, I'd need to be at a computer to remember the muscle memory though
16:56:15 <peter1138> Oh, does it? Didn't know that.
17:28:24 <_zephyris> It's got a lot more conventional its ui recently
17:28:59 <_zephyris> Classic blender is a great example of ultra streamlined and efficient ui, but really unconventional
17:30:11 <_zephyris> Kinda like factorio... It's like entering some kind of zen state.
17:36:32 <peter1138> Hmm... label of things that don't fit in the window. 1) make window bigger (urgh), 2) tooltip 3) something else...?
17:36:35 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
17:38:39 *** WormnestAndroid has joined #openttd
17:49:01 <xarick> `auto IsBetter = [&best] (const auto *current) {` CammelCase for lambdas?
18:25:00 <xarick> digitalfox: let me see which PRs I can live without
18:26:33 <xarick> #10411 - I can live without
18:26:33 <xarick> #10544 - I need it for my AI ships, current search for depot range is too short
18:27:29 <xarick> #12083 - I can live without
18:30:16 <xarick> #12084, #12087, #12088 - Well, if I only focus on vanilla, I can live without. But AIs can't really guess the properties of newgrfs airports
18:30:59 <xarick> #12093 - I can live without
18:32:46 <xarick> #12097 - I can live without, probably, not really using oilrigs atm
18:34:07 <xarick> #12249 - I'd like to try unbunch on my AI ships
18:36:15 <xarick> #12977, #13067, #13264, #13366, #13370, #13374 - I can live without, even though I wasted ton of time trying to improve rivergen
18:37:46 <xarick> #13398 - Well, I don't plan on using road waypoints, but... scripts currently don't support it properly
18:38:33 <xarick> #13419 - Same as above
18:39:56 <xarick> #13476 - I need it for my AI when using wallclock time mode
18:40:46 <xarick> #13565 - It would be really cool to have a way to setup different versions of the same AI, but I guess I could live without it
18:41:46 <digitalfox> Xarick I’m no one here, to say what PR should or shouldn’t be open.
18:41:46 <digitalfox> I just think in 115 open, you have 28 open, some from months, years ago.
18:41:46 <digitalfox> Maybe if the developers with power to aprove haven’t approved for so long, it might be because they are really busy, they don’t really see benefits, your code just doesn’t make the cut…
18:47:33 *** Wolf01 is now known as Guest11506
18:53:41 *** Guest11506 has quit IRC (Ping timeout: 480 seconds)
18:54:25 <ahyangyi> I think that having old pull requests is normal, because while it looks weird, it is often better than other alternatives, such as closing them.
18:55:22 <ahyangyi> Or keep track of those things in other ways
19:31:41 <Rubidium> digitalfox: for me it's mostly an issue of not listening to advice, getting annoyed by that and then just blocking it out to do the more fun things
19:36:47 *** gelignite has joined #openttd
19:41:33 <michi_cc> Rubidium: Something like "Fix: Alignment of 'i' with diacritics in OpenTTD Sans font"? (I don't think we gave Zephyris merge permissions yet 🙂
19:49:09 <_zephyris> Just saw this 🙂 Went for "Fix: i circumflex width in TrueType small font"
19:49:43 <_zephyris> (Noone noticed/reported the problem with rarer i diacritics yet)
20:25:32 <Rubidium> either message's fine by me
20:49:05 <xarick> I get ERR_ALREADY_BUILT when i build a rail track of a different rail type on a rail track
20:55:51 *** gelignite has quit IRC (Quit: Stay safe!)
21:02:11 *** tokai|noir has joined #openttd
21:02:11 *** ChanServ sets mode: +v tokai|noir
21:08:13 <andythenorth> but that was Chrome 😛
21:09:11 *** tokai has quit IRC (Ping timeout: 480 seconds)
21:14:16 *** nielsm has quit IRC (Ping timeout: 480 seconds)
21:17:27 <_zephyris> I got `ERR_ON_THE_SIDE_OF_CAUTION`, but that's just good advice.
21:30:29 <andythenorth> it's been a while 🙂
21:37:34 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
22:04:46 <locosage> KuhnovicviaGitHub: hm, something like this would be quite fitting for cmclient
22:07:25 <xarick> so i want blue to move back one piece and pf again
22:15:46 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
22:20:36 *** kuka_lie has quit IRC (Quit: Lost terminal)
23:00:39 <peter1138> Hmm, either I add a load of `std::move()`, or just copy the encoded string :S
23:01:15 <peter1138> Why does this work?
23:01:19 <peter1138> void CommandCost::AddCost(const CommandCost &ret)
23:01:24 <peter1138> this->encoded_message = std::move(ret.encoded_message);
23:02:13 <_jgr_> The std::move will just do nothing there
23:05:09 <frosch123> std::move is just a cast. In this case it will cast to `const string &&`. Move assignment does not match the const, so overload resolution matches the copy assignment
23:13:13 <peter1138> Okay, can I make the function accept CommandCost &ret and CommandCost &&ret... :S
23:15:06 <frosch123> Either you make it a template. Or you make it a plain value parameter
23:15:59 <frosch123> The latter results in one copy one move for const &, and zero copy two moves for &&
23:16:38 <frosch123> So one move more expensive than the optimum
23:17:32 <frosch123> Which may be preferable over moving the function into the header as template
23:20:12 <_jgr_> Even then, the compiler may well optimise that away for LTO release builds
23:28:24 <frosch123> I think there was a meme making fun of "C++03: pass stuff as const ref, it's the best" vs "C++11: pass stuff by value, it's the best"
continue to next day ⏵
