IRC logs for #openttd on OFTC at 2022-12-09
⏴ go to previous day
00:04:15 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
00:08:33 <Pruple> I'm still wondering how withdrawal works... I'm not sure keeping the parent in buy lists when it's no longer available makes sense. reparenting to the oldest/lowest ID currently available? and/or synchronised withdrawal for variants with the same intro date and lifespan?
00:10:29 <petern> In theory if reliability is synced, and the intro dates/lifespan are the same, it'll be synced withdrawal too. Perhaps.
00:17:32 <petern> Okay, Iron Horse is a good test of getting the order correct for reliability
00:51:34 <petern> Well it's not quite syncing withdrawal, but the top one is how the expired parent appears since a couple of days ago.
01:17:52 <Eddi|zuHause> i haven't followed the discussion, but: reliability syncing works on initial intro date, can't you just override the per-company mask to change variant availability?
01:20:17 <Eddi|zuHause> we might have shuffled that around in the past, but on model initialisation there's code to store/load random seed for the reliability curve
01:29:11 <petern> reliability isn't synced
01:34:29 <Eddi|zuHause> right, so someone undid this
01:34:57 <Eddi|zuHause> but it could technically be done this way
01:36:43 <petern> However that does rely on the intro date being the same.
01:44:10 <petern> Ah, and of course the variant parent is not flagged, so has different reliability anyway. Poot.
01:56:04 <petern> Urgh, andy's still got random-colour liveries hiding behind the variants 🙂
01:57:05 <petern> Confused me somewhat 😄
01:59:07 <petern> Anyway I can sync reliability by fixing the random seed and including the variant id (if sync is turned on) or just the engine id (if it's not)
01:59:34 <petern> That way it uses the same ID as for the parent, even if the parent isn't set to sync
02:00:03 <petern> reset_engines no longer randomises reliability though
02:02:11 <Pruple> I don't think that's a problem
02:17:21 <petern> Only that I found it handy to randomly test various engine expiry combinations 🙂
02:19:52 <petern> Hmm, so reliability can then be synced purely by using spec-state, and doesn't need state of other engines. Good.
02:20:10 <petern> Although this way won't work of the intro date is different.
02:27:41 <petern> Hm, perhaps I could use the variant parent's age in that case.
02:29:23 <petern> That would make them expire at the same time too (given the same base life), even if the intro date was later.
02:35:32 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
03:06:58 <petern> Oh, if the intro date is before the current game date, it calculated the age in months, by dividing the number of days by 32.
03:13:22 <Eddi|zuHause> that sounds almost correct :p
03:13:45 <Eddi|zuHause> maybe file that under 1mph=1.6kph :p
04:01:40 *** D-HUND has quit IRC (Ping timeout: 480 seconds)
04:21:38 <EmperorJake> Does the variants feature work for wagons as well?
07:33:17 *** sla_ro|master has joined #openttd
08:17:39 <andythenorth> petern: surprise!
08:25:58 <andythenorth> goes it variants today?
08:26:06 <andythenorth> I'm supposed to be working a half day
08:59:49 <petern> EmperorJake: It should do but I've not tried it
09:01:43 <EmperorJake> Good, because I can think of quite a few use-cases for it already in my GRFs 🙂
09:45:44 <petern> Those are probably the reason for it 😉
09:46:26 <petern> andythenorth: Goes it throw out colour-based trolling?
10:12:41 <andythenorth> I'll turn that off later
10:13:09 <andythenorth> I want to port the existing automagical CC liveries to variants directly
10:13:18 <andythenorth> instead of pissing around adding lots of new trains
10:13:38 <andythenorth> compile can sort that out, just requires attention / headspace 😛
10:39:05 <petern> Is there a reason to save/restore random seed state instead of just using a new Randomizer?
10:42:14 <petern> `Random()` has specialness when `RANDOM_DEBUG` is enabled I suppose.
10:45:02 <petern> Eddi|zuHause: In true old-code fashion, it's a cryptic `>> 5` rather than `/ 32`
10:46:30 <dwfreed> which makes no sense, because compilers will automatically optimize a / 32 to a >> 5 in the assembly
10:53:37 <petern> ConvertDateToYMD is complex 🙂
10:59:33 <petern> So if the age is calculated 10 years after the intro date, it's a 6 month difference. Does it matter? Probably not.
11:00:05 <Timberwolf> Good old Chris Sawyer. Never do a simple calculation when you can shift, twiddle a couple of bits and shift again to save 2 cycles.
11:00:30 <Timberwolf> Even if it's something only called once per frame!
11:02:12 <petern> That isn't quite fair.
11:02:19 <petern> CS was writing in assembly.
11:04:50 <Timberwolf> There's something somewhere where he admits it's habit, having to do it so often for tight loops in simple games that it's just habit even in places where a MUL or DIV with an extra memory access would be fine.
11:06:08 <petern> Even in 1995, every instruction counted.
11:14:50 <petern> Probably less so when he wrote Locomotion 🙂
11:15:21 <petern> Anyway a compiler will (probably) compile /32 as >>5.
11:21:58 <LordAro> until you turn optimisations on
11:23:24 <Timberwolf> RCT is the one I tend to think of where he does it for things like the ride rating calculation.
11:24:18 <Timberwolf> But I think a lot of it is having written so much 8-bit stuff and conversions to PC when they were very limited, it's just ingrained by that point.
11:24:41 <Timberwolf> And it's not like it hurts if you're the only person reading it and to you that's the "natural way" of, say, dividing by 5/8ths.
11:27:00 <Timberwolf> There's a good thing from Carmack somewhere where he talks about doing "dumb things really fast in assembly" in Wolfenstein to using smarter algorithms in Doom and writing pretty much the entire thing in C apart from a few of the *really* tight loops like the wall column renderer.
11:39:57 <petern> Never forget the Fast inverse square root algorithm from Quake 🙂
13:13:36 <petern> Hmm tempted to start using flags fields in databases, instead of lots of individual booleans.
13:14:00 <petern> Bad for querying and normalization, good for quickly adding something without updating schema
13:23:14 <TallTyler> I strongly considered flags for employee qualifications in my current Blazor application, but two-way binding is so much easier with booleans that I opted for faster development time and a slightly less optimized database
13:24:13 <TallTyler> (Binding with the model object, not directly with the database, of course)
13:24:33 <petern> I use some views to simplify getting the right data without having code for it. So I need to update the real object, and the views, and... meh.
14:27:36 <petern> And then you think maybe just a flag isn't enough information anyway...
14:33:28 <petern> And is it too hard to set up if I make it flexible enough.
14:54:30 <petern> Crap, now I need to run reset_engines twice. Maybe I should 😄
14:55:08 <petern> Ah no, this is still wrong.
15:07:51 *** WormnestAndroid has joined #openttd
16:02:36 *** gelignite has joined #openttd
16:09:00 *** sla_ro|master has quit IRC ()
17:32:31 *** HerzogDeXtEr has joined #openttd
18:20:59 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
18:25:42 *** sla_ro|master has joined #openttd
18:43:34 <DorpsGek> - Update: Translations from eints (by translators)
18:48:01 *** Wormnest has joined #openttd
19:18:04 <petern> Well... I wonder what actually happens with reliability syncing with different introduction dates.
19:18:37 <supermop_toil> different reliability per livery?
19:26:38 <petern> That's why I'm syncing it, so it's the same.
19:26:52 <petern> It now works if the intro date is the same, which is a start.
19:27:03 <petern> It should also work even if it isn't, but I'm not sure 🙂
20:00:52 <andythenorth> petern: no new variants, but nerfed the CC stuff
20:06:55 <andythenorth> testing something about ID ranges here 😛
20:07:03 <andythenorth> that's one engine 😛
20:10:30 <petern> Are you using 64k engine IDs yet? 😄
20:11:44 <andythenorth> the IDs go in blocks of 10
20:11:57 <andythenorth> I am using the intermediates for variants 😛
20:12:14 <petern> Hmm, just turned my boiler temperature up, it's cold enough that 60 isn't warming the house up.
20:12:41 <andythenorth> resilient: 13980, 13981, 13982 etc
20:13:12 <bigyihsuan> only 10 variants? 😛
20:13:22 <andythenorth> they can exceed 10 😛
20:13:29 <petern> So, extra flags... I only need 4. Should I make it a DWORD anyway, so there's enough room for future flags without adding more properties...?
20:13:32 <andythenorth> just the next ID has to be left unused by anything else
20:13:55 <andythenorth> what do other flags have?
20:14:34 <petern> The first set of flags is 8 bits.
20:15:00 <petern> The other way to extend that is to create NewGRF v9. But no.
20:15:27 <petern> DWord is overkill but means there's future space for lots of things even though I can't think of anything right now.
20:17:29 <andythenorth> flags for air brake or vacuum brake or both 😛
20:17:43 <andythenorth> oh no, we'd need probably actual PSI and line charge speeds
20:17:49 <andythenorth> for true fidelity
20:17:53 <andythenorth> and gold plated AC cables
20:27:09 *** reldred has quit IRC (Quit: User went offline on Discord a while ago)
20:53:18 *** Wormnest has quit IRC (Read error: Connection reset by peer)
20:53:38 *** Wormnest has joined #openttd
21:29:35 <petern> Time to throw that cheese out
21:30:54 <petern> I should update the PR
21:31:04 <petern> Oh yes, when it compiles.
21:41:03 <petern> Thanks, I grew it myself
21:41:45 <andythenorth> consist.add_variant(
21:41:58 <Pruple> you can pick your friends, but you can't pick your nose
21:42:20 <Pruple> variants will be handy for road vehicles with extra trailers 🙂
21:42:43 <Pruple> times I have completely rewritten the plan for my road vehicle set since the last time I actually drew a road vehicle sprite: 4.
21:43:44 <andythenorth> variants with 1, 2, 3, 9 trailers?
21:54:43 <petern> I need to pop out, I expect someone to have tried using these flags and reporting bugs 😉
21:58:07 <andythenorth> Pruple: will do it 😛
22:15:51 <andythenorth> ok brains, need help 😛
22:17:10 <petern> There's a tech tree? Heh
22:18:02 <petern> I came back in, I was not needed for lift duties after all
22:19:11 <petern> You could make it fold/unfold lol
22:20:17 *** wallabra has joined #openttd
22:20:55 *** wallabra has joined #openttd
22:23:41 *** wallabra has joined #openttd
22:37:11 *** keikoz has quit IRC (Ping timeout: 480 seconds)
22:49:00 <andythenorth> oops fell asleep on the sofa
22:52:46 <andythenorth> ok tomorrow, more variants 😛
22:53:05 <petern> lol, when you name-drop OpenTTD and they're like, what, never heard of it.
22:57:39 *** Tirili has quit IRC (Remote host closed the connection)
23:04:28 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
23:06:34 *** Samu has quit IRC (Quit: Leaving)
23:10:56 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
23:18:13 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
23:18:29 *** WormnestAndroid has joined #openttd
23:23:41 *** gelignite has quit IRC (Quit: Stay safe!)
23:26:32 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
23:27:13 *** WormnestAndroid has joined #openttd
23:32:35 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
23:37:41 *** sla_ro|master has quit IRC ()
23:48:15 *** Wormnest has joined #openttd
continue to next day ⏵
