IRC logs for #openttd on OFTC at 2025-11-07
⏴ go to previous day
00:27:53 *** goddess_ishtar has joined #openttd
00:27:53 <goddess_ishtar> kaji_kaede: tbh I just clone a git repository and use Visual Studio Code as my text editor
00:28:00 <goddess_ishtar> there really isn't much to it
00:28:23 <goddess_ishtar> whenever I want to build my branch I just use cmake
00:29:39 <goddess_ishtar> yeah VSCode is a little barebones but frankly all of the IDEs I've used are just more bloated for the same features
00:30:36 <goddess_ishtar> at least VSCode *works* and doesn't have a bunch of options I need to mess with first and menus on top of menus on top of menus
00:31:56 <goddess_ishtar> I've tried to use Code::Blocks in the past, and my install would just break for random reasons and trying to use solution files was a nightmare
00:37:00 *** Zathras_4 has joined #openttd
02:21:54 *** Wormnest has quit IRC (Quit: Leaving)
03:51:59 *** gnu_jj_ has quit IRC (Ping timeout: 480 seconds)
04:30:28 *** Zathras_11 has joined #openttd
04:30:31 *** Zathras_1 has joined #openttd
04:33:51 *** Zathras_4 has quit IRC (Ping timeout: 480 seconds)
04:34:01 *** Zathras has quit IRC (Ping timeout: 480 seconds)
04:39:09 <DorpsGek> - Update: Translations from eints (by translators)
05:36:28 *** greeter has quit IRC (Remote host closed the connection)
05:41:01 *** greeter has joined #openttd
05:57:19 *** tokai|noir has quit IRC (Ping timeout: 480 seconds)
06:14:10 *** ChanServ sets mode: +v tokai
06:43:00 *** tokai|noir has joined #openttd
06:43:00 *** ChanServ sets mode: +v tokai|noir
06:43:44 *** tokai has quit IRC (Ping timeout: 480 seconds)
07:00:59 *** WormnestAndroid has quit IRC (Remote host closed the connection)
07:01:04 *** WormnestAndroid has joined #openttd
08:53:12 <pickpacket> Everyone good in here? It's Friday, peeps
09:07:10 <pickpacket> LordAro: no work today?
09:07:25 <LordAro> i suppose i should at some point
09:26:56 *** orudge` has joined #openttd
09:26:56 *** ChanServ sets mode: +o orudge`
10:15:18 *** qushev_ has joined #openttd
10:16:19 <qushev_> Has anyone installed OpenTTD on a VDS server?
10:16:41 <qushev_> I have a problem with the openttd.cfg file. I can't add the server_name and server_password parameters. Has anyone solved this problem?
10:18:13 <LordAro> server_password no longer exists
10:18:16 <LordAro> server_name goes in private.cfg
10:18:34 <LordAro> (look at server_game_type instead of the password)
10:21:00 <LordAro> also what on earth is a 'virtual dedicated server'. Those two terms are opposites
10:22:48 <Borg> hah, someone came up w/ new slogan... to bump up their sales...
10:23:32 <LordAro> wait, i'm very much wrong about server_password
10:23:36 <LordAro> that's in secrets.cfg
10:23:50 <LordAro> our documentation for this really sucks
10:24:17 <reldred> I mean if they mean a VPS on a host they haven’t oversubscribed the everloving fuck out of that’s a useful distinction 🤣
10:25:11 <reldred> Love how cheap my ssdnodes VPS is but sometimes it runs like an absolute dog. Win some lose some.
10:25:54 <reldred> Yeah that ‘if’ wasn’t just doing some work it was doing overtime
10:28:09 <mmtunligit> im trying to hide a widget the proper way as requested, but i cant seem to figure out why it isnt working, i think im doing everything correctly. right at the start of the struct for the window after initializing the variables i have
10:28:09 <mmtunligit> ```void HideButtons()
10:28:09 <mmtunligit> bool test = false;
10:28:09 <mmtunligit> this->GetWidget<NWidgetStacked>(WID_QS_MOVE_SEL)->SetDisplayedPlane(test ? 0 : SZSP_NONE);
10:28:10 <mmtunligit> and then down in the widget parts i have
10:28:12 <mmtunligit> ```NWidget(NWID_SELECTION, INVALID_COLOUR, WID_QS_MOVE_SEL),
10:28:12 <mmtunligit> NWidget(WWT_TEXTBTN, COLOUR_GREY, WID_QS_MOVE), SetMinimalSize(65, 12), SetFill(1, 1), SetStringTip(STR_BUTTON_MOVE),
10:28:14 <mmtunligit> EndContainer(),```
10:28:14 <mmtunligit> but it still shows the move button every time and i cant figure out why
10:34:16 <mmtunligit> wait hold on, i might be dumb
10:48:42 <xarick> how do i click fast forward with the console
10:53:39 <xarick> nevermind, there's still graphics
10:56:43 <xarick> gonna try following this
11:10:15 <Borg> damn.. those new LinkGraphs colors + new logic for manual cargo dist kicks ass
11:10:41 <Borg> now you can see where you have underutilized paths or where you can get more cargo if needed
11:11:29 <Borg> until now I used LinkGraph only to see where I actualy haul what.. thats all.. because LinkGraph itself doesnt carry any usefull info for DT_MANUAL..
11:22:18 *** qushev_ has quit IRC (Quit: Page closed)
11:23:05 <xarick> I don't know how many ticks i need
11:46:56 <xarick> 30000 ticks was insufficient
11:54:44 <xarick> how many ticks in 3 years?
12:03:33 <xarick> interesting... the game is paused but in fast forward
12:13:41 *** michi_cc has joined #openttd
12:13:41 *** ChanServ sets mode: +v michi_cc
12:36:38 *** rito12_51026 has joined #openttd
12:36:38 <rito12_51026> Why the PR updates didn't show up on discord since yesterday? They appear in the IRC logs, however.
12:44:59 <mmtunligit> yeah, i noticed that too
12:47:09 *** ialokin has quit IRC (Ping timeout: 480 seconds)
12:51:08 <_glx_> xarick: GS keep running while game is paused, because they can control the pause too
12:52:59 <mmtunligit> Discord Moderator: github bot seems to have broken
13:00:01 <_glx_> They can't do anything about it
13:07:18 *** olionkey has joined #openttd
13:09:30 *** ialokin has joined #openttd
14:17:22 <_glx_> hmm but IRC received them
14:37:38 *** Flygon has quit IRC (Read error: Connection reset by peer)
14:39:53 <_jgr_> Maybe, but I got tired of messing about with ZNC just to get scrollback that sort of half works 😛
14:42:00 <LordAro> permanent tmux session with irssi :p
14:42:08 <LordAro> (and also znc on top of that)
14:42:21 <LordAro> sometimes scrollback works
14:42:37 <LordAro> lately my username highlighting has been broken, it gets recoloured every time i reconnect to the tmux. not sure why
14:45:08 <Borg> _jgr_: hmm never used ZNC.. back in old days I wrote my own simple BNC that did buffered everything and syncing back to client on connected.. worked kinda flawlessly
14:45:17 <Borg> but then I moved most of my stuff to CLI. so irssi + screen
15:24:59 * Rubidium wonders who remembers the fun of using BNC (connectors) for ethernet
15:29:14 <_glx_> and the termination on last T ?
15:33:11 <peter1138> AUI-to-BNC transceivers
15:35:30 <peter1138> There was also that place I went to that had thicknet -- 10base5
15:36:59 <peter1138> (Although at that point it may just have been left there and not actually in use, it was early 2000)
15:42:11 <peter1138> Crap, some slop-machine is crawling my gitea instance :(
15:43:03 <Borg> Rubidium: I had 10Base2 network w/ friends in mid 90's :)
15:43:45 <Borg> so Ive done quite a few BNC connectors.. and yeah, I had to keep a bounch of T like splitters + terminators.. 50Ohm
15:44:22 <Borg> peter1138: filter them!!!!! ;)
15:46:17 <peter1138> Back in the mid 90s I used a serial cable to connect two 486 PCs to play Doom.
15:46:25 <peter1138> For some reason I still have the serial cable.
15:47:28 <Borg> yeah. we did that too.. before LAN
15:48:22 <Borg> I actually.. still play DOOM w/ friends sometimes.. albeit.. using source port :) and moded DOOM
15:51:51 <peter1138> Anyway, I'm using gitea's built in rate limiting but clearly that isn't enough.
15:53:03 <Borg> yeah.. its harder and harder.. every year
15:53:09 <Borg> I moved out of internet most of my services..
15:53:18 <Borg> kept just skeletonal web pages + MTA
15:53:24 <Borg> everything else went to VPN overlay network
15:53:51 <Borg> MTA actually too... 2 frontends + 1 backend setup..
15:56:12 *** f_ is now known as funderscore
15:56:27 *** funderscore is now known as f_
16:04:33 *** f_ is now known as funderscore
16:10:21 *** efessel has joined #openttd
16:10:21 <efessel> a null modem cable they called it right?
16:22:37 *** Wormnest has joined #openttd
16:44:05 *** orudge` has quit IRC (Quit: Goodbye.)
17:21:18 <xarick> is discord pdf friendly?
17:24:13 <xarick> benchmarks compiled into a table
17:27:02 *** gelignite has joined #openttd
17:52:07 *** audigex has joined #openttd
17:52:07 <audigex> I assume this has been reported as a bug with vanilla industries and presumably rejected from being fixed - does anyone know where it was discussed? I know there's a "that's how it's always worked" argument for vanilla industries, but that 100% seems like it should be fixed. It's absurdly unintuitive
18:01:41 *** talltyler has joined #openttd
18:17:45 <andythenorth> Pitchforks though?
18:17:59 <peter1138> Don't fix non-bugs, eh?
18:19:20 <LordAro> might be worth documenting it properly though
18:21:21 <LordAro> it is from the perspective of users who might go looking there for it
18:22:33 <LordAro> covers half the issues in that document, looking at it
18:23:06 <LordAro> and the other half can probably be removed ("SDL in Ubuntu 9.04 may have excessive CPU usage")
18:24:55 <_jgr_> I'd expect that very few users are even aware that that document exists
18:39:51 <Borg> very annoying.. thats why BSPI adjust tiles of all industries
18:46:42 <peter1138> That's why NewGRF lets you do that.
18:47:20 <Borg> yeah.. maybe someone should provide vanillia industries NewGRF. w/ fixed stuff?
18:47:33 <audigex> peter1138: I’d 100% consider that a bug, personally. The UI clearly doesn’t match the result
18:47:33 <audigex> I’m all for backwards compatibility but I can’t imagine fixing that is actually gonna break many people’s save games
18:47:49 <peter1138> Someone could make a vanilla-style industry set that makes all tiles have acceptance, if they wanted to.
18:47:59 <audigex> Borg: Also fix the lack of never/expires helicopters
18:48:26 <Borg> audigex: im suprised there is no NewGRF that does that already
18:48:44 <audigex> peter1138: If it could be included in the game and used by default for new save games or something, that would be good
18:48:44 <audigex> But if it’s just a “you have to go find it” then it’s not even vaguely going to fix anything for new users who’d never know to go looking for it
18:49:22 <audigex> Borg: As above, I’ve thought about doing it but unless it’s used by default it’s a waste of effort - the type of user who is experiencing these bugs isn’t going to go find the “fixes bugs” newGRFs
18:49:42 <audigex> If we include it by default in new games I’ll happily make a “community fixes” newGRF
18:49:45 <Borg> do stuff for yourself.. and then share
18:49:49 <Borg> not the other way around
18:50:15 <Borg> I did BSPI because I wanted... when it was in ready shape.. I shared.. and I didnt give a flying f... whatever it will be liked or not..
18:50:21 <_jgr_> If you're going to install a GRF to avoid problems in the vanilla vehicles, you might as well go with one of the many GRFs which provide a whole suite of different vehicles
18:50:58 <Borg> if I would be maintainer of OpenTTD I would definitly fix it
18:51:12 <Borg> Im that cargo acceptance of industry tiles are overlook
18:51:23 <audigex> Borg: I don’t need it for myself, use FIRS. If I did it it would solely be for others, and I’m not going to waste that effort on something that, as JGR says, doesn’t really make sense to have as an optional newGRF the user has to go find
18:51:25 <Borg> cost is minimal.. and I cant image it can break anything
18:52:47 <mmtunligit> i think if fractional cargo acceptance were simply explained *anywhere* that didnt require explicitly looking for it it wouldnt be an issue
18:53:07 <Borg> mmtunligit: but its not a problem about fractional cargo acceptance here
18:53:12 <audigex> “Make it for myself, release it in case others like it” is 99% of my approach to newGRF development (I maintain the Thomas the Tank Engine set…), but in this instance if it’s not included by default it’s literally a waste of time
18:53:12 <audigex> I’d prefer it to be fixed in the base game, but the only other solution I can see if the devs didn’t want to do that (because of backwards compatibility with save games) would be to have a default included newGRF
18:53:17 <Borg> some tiles have 8/8 and some none.. with is.. weird
18:53:31 <xarick> I'm too dumb for git bash
18:53:38 <audigex> mmtunligit: This isn’t about fractional acceptance - it’s about the UI showing an industry in the acceptance area but it isn’t actually accepted
18:55:20 <mmtunligit> yeah but thats due to the same sort of thing that leads peopel to get confused about it in towns too, some tiles, despite looking like they should accept cargo, dont. in the case of houses its that there arent enough, and for industries its that those tiles just dont accept any, but ultimately this is the same problem
18:59:18 <mmtunligit> unfortunately theres no meechanism i know of to have a popup the first time someone runs into this, and then never again
18:59:25 <peter1138> If only there was some kind of display on screen that tells you want will be accepted and supplied when building a station...
19:00:32 <mmtunligit> ok but players are dumb and do not always look at that, they look at the tiles on the ground and if it looks like it covers the industry they will assume it should work
19:02:46 <mmtunligit> idk its probably fine the way it is, but it is confusing to newer players (and sometimes even older ones, speaking from experience)
19:06:38 <rito12_51026> Mayby just hide the tiles that do not work in such situation?
19:07:54 <audigex> peter1138: Sure, but that’s a level of detail a new player is going to miss
19:07:54 <audigex> And what’s the point of a coverage area if it’s not accurate? Clearly the displayed coverage area should represent…. Ugh, coverage
19:08:05 <mmtunligit> no, thats not an actual explanation
19:08:16 <mmtunligit> (directed at rito)
19:09:44 <audigex> The current situation is absolutely a bug
19:09:44 <audigex> The question is whether it gets fixed directly (potentially breaking savegames) or indirectly (with a newGRF automatically included with newly started saves or something)
19:10:25 <audigex> If the interface shows that a station will cover the industry and it doesn’t, that’s a bug and I don’t see how anyone can say otherwise even if a different part of the UI does reflect the situation properly
19:10:36 <peter1138> It's not a bug, so it's not going to be changed in the code.
19:10:38 <mmtunligit> possibly could be solved with a message like "need more industry/house tiles in coverage area for acceptance/supply of [CARGO]" (depending on the situaion) its a bit of a lie but it will solve the problem
19:10:42 <peter1138> You can change it via NewGRF.
19:10:46 <audigex> peter1138: Of course it’s a bug
19:11:44 <mmtunligit> yeah this isnt a bug, my understanding the industries were designed with some tiles not accepting the cargo (why would smokestacks be a valid place to drop off coal?)
19:12:10 <talltyler> Bug = not intended by the developer. Chris Sawyer definitely did this on purpose, so not a bug.
19:12:10 <talltyler> Bad game design? I would argue yes. Even though it’s not a bug.
19:12:14 <audigex> Sorry but that’s just plain silly. There’s no world in which the UI showing something that objectively isn’t the case, isn’t a bug
19:12:14 <audigex> The reason for the bug (badly designed vanilla industries) doesn’t change the fact it’s clearly wrong now
19:12:14 <audigex> The UI absolutely shows that the station coverage area reaches that industry
19:12:39 <audigex> Or are we seriously going with a pedantic “technically intended 30 years ago” here?
19:13:06 <mmtunligit> yeah this isnt a bug, its badly explained mechanics
19:13:25 <mmtunligit> the coverage area doesnt shwo acceptance, it shows the potential for acceptance
19:14:07 <rito12_51026> talltyler: Why he didn't do it by mistake, like pressed wrong button or sth?
19:14:29 <audigex> Okay, my apologies for the word but.
19:14:29 <audigex> Correction; it’s a fucking awful deliberate feature from 30 years ago that needs to be changed due to confusing players in a way that is completely and utterly indistinguishable from a bug unless you know that it was done deliberately
19:14:29 <audigex> Phrase it however you want, the mechanic is shit and should be fixed if it’s a bug, or improved if it’s not
19:15:04 <peter1138> The mechanic can be fixed.
19:15:52 <audigex> peter1138: Which does NOTHING for new players who would never find it
19:15:52 <audigex> Unless the newGRF is included with newly started saves. Which is what I was asking about
19:15:58 <mmtunligit> being changed isnt something thats going to happen on account of the "remain faithful to the origional game" part of the mission statement, the "fix" here is as peter says, a newgrf, or better messaging, which i think i had a pretty good idea for a bit further up but id like to hear other people opinions
19:16:36 <audigex> Fixing it with a newGRF without including it in new saves is a complete waste of time and does nothing to “fix” the bad mechanic for the people it actually affects
19:17:18 <audigex> The only other thing I could see helping would be if the coverage area highlighted green/red based on which tiles accept/don’t accept cargo or something
19:17:21 *** michi_cc[d] has joined #openttd
19:17:21 <michi_cc[d]> Well, we added more refits to the default vehicles without any NewGRFs too, so changing the industry acceptance directly wound not be any different.
19:17:55 <michi_cc[d]> It is of course still a behaviour change, so the question of "faithful" still remains.
19:17:56 <rito12_51026> audigex: An included tutorial could fix this by explaining that it is a deliberate feature and can be modified with use of NewGRF
19:17:59 <audigex> michi_cc[d]: Seems sensible. Same with the helicopter expiring
19:17:59 <audigex> “It’s not in the original TTD” just seems wildly bloody-minded as a reason not to improve things
19:18:06 <audigex> rito12_51026: Or we could just change it.
19:18:16 <vondpc> i feel you always will face 2 different camps
19:18:59 <audigex> “This terrible feature exists and we’ll tell you about it in a tutorial you may or may not use and probably won’t remember” is a much worse solution than just changing something that nobody will actually be negatively affected by if it’s changed
19:19:07 <talltyler> Expecting players to read is a fool’s errand. I do like the idea of showing acceptance on the coverage highlights, but that’s still not super obvious.
19:19:23 <talltyler> “The tutorial covers it” is not a solution 🙂
19:19:40 <audigex> talltyler: Yeah it’s my 3rd preferred option, I suggest it simply because the other options seem to be meeting some resistance for some reason
19:19:44 <talltyler> (Especially because we don’t even have an official tutorial)
19:19:49 <mmtunligit> yeah i suppose more words isnt much of an answer when not but a few minutes before i did the "players dont read bit lol"
19:19:51 <vondpc> you could ay the entire block of tiles belongs to the industry and having coverage that overlaps with any part of it should be considered as accpeted
19:20:13 <talltyler> vondpc: That’s what my linked PR proposed to do 🙂
19:20:27 <talltyler> Without changing GRF industry tile inheritance, even
19:21:51 <audigex> talltyler: Yeah that seems like a sensible fix (improvement, whatever)
19:22:29 <mmtunligit> yeah, coverage ighlight changing dynamically based on acceptance might be best, red for no acceptance, green for acceptance, maybe yellow for partial acceptance/supply? not sure how thatd work with multiple overlapping factors on one tile though
19:22:52 <mmtunligit> and accesebility, though i dotn knwp how seriosuly we take things like a colorblind mode as-is
19:23:08 <kaji_kaede> audigex: Just create a fork that overtakes vanilla OpenTTD in popularity and get it on Steam so that new players come to your version first.
19:23:49 <mmtunligit> kaji_kaede: "quick" way to get atari on your ass too
19:23:55 <rito12_51026> How about reversing the coverage display so it is around the industry instead of station and the player has to place station tiles inside the area.
19:23:58 <talltyler> My PR would make it easy to make a GRF with original partial acceptance (literally just redefine the industries and inherit all the properties), while vanilla industries get full inheritance.
19:24:40 <mmtunligit> rito12_51026: i think youd have to make it a toggle
19:24:44 <talltyler> rito12_51026: Wouldn’t work if you have multiple industries near each other, also doesn’t show you houses
19:25:14 <mmtunligit> and roadstops have smaller acceptance radius iirc
19:25:23 <talltyler> mmtunligit: We have colour scale settings for the linkgraph, I don’t think we should go backwards on colour blind friendliness 🙂
19:26:00 <talltyler> Well, just the one setting actually
19:26:06 <kaji_kaede> mmtunligit: Yeah. Just get sued. It's that easy.
19:26:09 <mmtunligit> no i dont think we should either, i was just wondering about how introducing color coding tile highlights would work with that
19:26:41 <mmtunligit> talltyler: well if you code it like town coverage and station coverage you cant have multiple up at once
19:27:09 <talltyler> I’d rather just change the confusing behavior than try to explain it better 🙂
19:27:11 <mmtunligit> selecting a station to build would also deslect the industry
19:27:25 <rito12_51026> talltyler: Each industry could have different colour and size of the box
19:28:03 <talltyler> Sounds like a mess tbh
19:28:09 <mmtunligit> no just make it one-at-a-time like every other coverage highlight
19:28:17 <mmtunligit> we have a pattern, we should stick to it
19:28:46 <talltyler> OpenTTD has enough complexity, we need some “design by subtraction” 🙂
19:29:44 <rito12_51026> Just totally remove the display
19:30:10 <talltyler> (I don’t mean just taking stuff away, I’d consider depot unbunching a “design by subtraction” feature because it means you don’t have to faff around with timetables unless you really want to)
19:32:01 <mmtunligit> im slowly coming around to the "this should just work" camp, though i still think fractional acceptance/supply with houses would remain pretty confusing
19:32:22 <mmtunligit> you cant just say "all these are 8/8 now" without affecting the balance
19:32:33 <talltyler> What is “balance”? 😛
19:32:41 <rito12_51026> Or maybe replace the coverage with the lines connecting industries in the range with the station center or mouse?
19:33:01 <talltyler> Fractional house acceptance is also frequently confusing, although my PR doesn’t touch that
19:33:23 <rito12_51026> Or highlight industries in range
19:34:26 <rito12_51026> with a brighter palette like it is done for hiding
19:42:05 <peter1138> Hmm, either this wad is buggy, or dsda-doom is buggy... or I just suck at Doom.
19:45:04 <andythenorth> why don't people just use IFRS? :eyes?
19:45:24 <andythenorth> then industry acceptance is solved
19:46:58 <xarick> git bash cli is horribly unintuitive...
19:47:33 <xarick> how do i refresh the list of branches available on some remote?
19:47:46 <xarick> i can't do this simple stuff with the gui
19:52:03 <xarick> i don't see the "suspendable" branch
19:52:33 <peter1138> git fetch name-of-your-upstream
19:57:20 <xarick> thanks. there's a fetch in the drop down menu and it's gray though...
19:57:30 <xarick> have to resort to ugly cli
19:57:44 <peter1138> Don't be scared of CLI. It's how computers are meant to be used.
20:05:55 <andythenorth> it's how git is meant to be used
20:06:02 <andythenorth> don't trust git UI clients, they hide stuff
20:06:48 <andythenorth> can't draw pixels very well in the shell though
20:09:53 <peter1138> Oops, I built the Doom editor in debug mode :o
20:15:53 <peter1138> Okay, that's evil. This secret area has single-use switches to open up, and closes if you backtrack in a certain place.
20:16:22 <andythenorth> I haven't made a wad since probably 1997
20:19:25 <rito12_51026> If I change the order of the lines in the interactive rebase window (tmp text file edited in nano) will that change the commits' order or break something?
20:20:07 <peter1138> It changes the order.
20:23:37 <mmtunligit> if it runs into issues itll pause the rebase and let you decide what to do
20:23:52 <mmtunligit> i did it for the first time the other day, it was scary but i lived
20:40:51 <rito12_51026> Deleting is fun, went from +1,367 additions to only +256
20:45:56 <peter1138> Can be shorter I think.
20:52:24 <andythenorth> shall I do iron_horse.wad?
20:56:17 <andythenorth> lunch.wad first probably
21:02:02 <peter1138> He's so damn picky.
21:05:23 <andythenorth> hmm I never did the docs for grf stuff
21:05:33 <andythenorth> I have found all the open tabs of shame 😛
21:14:37 <peter1138> Urgh, never did buy a Pi 500+ for testing, just looked and the price increased £20 at some point :o
21:18:33 <andythenorth> is that an M4 Pi?
21:26:59 <peter1138> M5 is the new thing, no?
21:31:09 <xarick> _glx_: I'm impressed by the speed
21:32:44 <andythenorth> does M5 come after M4?
21:32:47 <andythenorth> or do I need both?
21:38:24 <_glx_> xarick: gamescript implementing its own valuate ?
21:38:50 <xarick> it's using the suspendable innate
21:39:25 <xarick> no custom implementation in gs
21:39:39 <_glx_> difference in ops seems strange
21:40:44 <xarick> might be because of item repeat?
21:40:44 <_glx_> because suspendable Valuate is just the original Valuate (with some changes) wraped inside a squirrel while loop
21:41:37 <_glx_> so I would expect a little more ops for suspendable than master
21:42:15 <_glx_> though it used more ticks, so it got suspended
21:42:56 <xarick> need to check what it does with items though
21:43:41 <xarick> unsure what resume_item implies...
21:44:35 <_glx_> the ops limit is now set to the remaining ops for the tick, while in master it's a huge ops limit
21:45:01 <_glx_> when the limit is reached the current item is stored so it can resume from that
21:45:36 <_glx_> and the squirrel while just calls internal Valuate until it's done
21:45:51 <xarick> is the evaluation still completed for that item that gets interrupted?
21:46:15 <_glx_> it may or may not, anyway it's discarded and redone on next call
21:46:36 <_glx_> that's why you see a change in result.txt
21:46:58 <xarick> aww... that's a problem, isn't it possible to get it complete then interrupt?
21:47:34 <_glx_> if I put a Sleep before this call in main.nut the original regression output is restored
21:48:30 <_glx_> interruption might happen anywhere during the valuation (especially if using a squirrel valuator)
21:49:27 <xarick> sleep before valuate? let me try
21:51:39 <xarick> in normal speed, sleep is bad
21:51:43 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
21:51:52 <_glx_> basically if you sleep before valuate you allow the full ops allowed for a tick
21:52:18 <_glx_> still less ops than what was allowed in master
21:53:10 <xarick> could it reset the ops counter before starting the valuate?
21:53:28 <peter1138> andythenorth, I would assume after?
21:53:42 *** Borg has quit IRC (Quit: leaving)
21:53:50 <peter1138> > With up to 3.5x more performance for AI workflows
21:54:05 <_glx_> the idea is to not use more ops per tick than allowed
21:54:14 <andythenorth> gotta have some AI to keep that stock price up
21:54:40 <_glx_> so stupidly huge lists don't affect actuall length of a tick
21:55:04 <andythenorth> if I could run a local diffusion model for reliable pixel art, I would
21:55:16 <andythenorth> but smush doesn't add anything
21:55:42 <_glx_> and as a bonus no more crashed script if they have too many items for Valuate
21:55:57 <peter1138> _glx_, is that a PR I should've reviewed?
21:56:25 <xarick> can't cheat the machine?
21:57:05 <_glx_> the more items in the list, the more ticks needed to run Valuate
21:57:19 <_glx_> without affecting game speed for user
21:58:41 <xarick> you say if you do a sleep, it will output the same as is in regression. So i though, do a pseudo-sleep before starting valuate, reset the current remaining ops to whatever is allowed for a tick before initializing the valuate
21:59:31 <_glx_> well the Valuate wrapper could for a sleep before the while loop, but it's not necessary
21:59:57 <_glx_> in regression example you see a diff only because valuator calls Random()
22:01:55 <peter1138> Hmm, did I imagine someone had made an extra highlight layer sprite?
22:05:25 <xarick> I like the concept, it's even as fast as master, but ... repeating the same item is... an issue.
22:06:36 <xarick> imagine someone enumerates the stuff
22:07:08 <xarick> uses the valuator function as adding +1 to a counter
22:07:25 <xarick> and at the end it's gonna be with a different result
22:08:02 <_glx_> that's abusing Valuate() 🙂
22:12:07 <xarick> I think i've seen some valuate functions that do some work on the side than just returning a value
22:12:13 <xarick> not sure who's the author
22:26:25 <peter1138> Does it start with "Xar" and end with "ick"?
22:28:57 <xarick> ``` function CustomValuatorFunction(item, instance)
22:28:57 <xarick> instance.valuate_counter++;
22:28:57 <xarick> return GSBase.ChanceItem(item, 1, 2);
22:28:58 <xarick> `list1[sorter_type][ascending].Valuate(CustomValuatorFunction, this);`
22:34:21 <peter1138> Hmm, what's more 'efficient', `x = list.emplace_back(); x.a = a; x.b = b; x.c = c;` or `list.emplace_back(a, b, c);`...
22:37:54 <LordAro> i'd have thought the latter
22:38:11 <LordAro> though there's a decent chance the compiler optimising it to be the same anyway
22:55:06 <peter1138> Yeah, depends on the compiler for sure.
22:57:21 <peter1138> Different code for each though.
23:01:40 <audigex> I've proposed re-opening talltyler's PR here, although would welcome discussion of the other options I've outlined (or any other options/views) on Github where the conversation can be more permanent than Discord's fleeting interactions
23:13:16 *** keikoz has quit IRC (Ping timeout: 480 seconds)
23:19:28 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
23:41:09 <_glx_> xarick: it's strange, there's only 1 tick difference meaning only 1 call got suspended and an item got valuated twice
23:42:12 <_glx_> hmm unless they all got suspended and spread over just one tick more
23:42:57 <talltyler> audigex: Re-opened for you. 🙂
23:47:36 <_glx_> anyway valuators are supposed to set the value, not have side effects
continue to next day ⏵
