IRC logs for #openttd on OFTC at 2023-05-07
⏴ go to previous day
01:03:33 *** WormnestAndroid has quit IRC (Remote host closed the connection)
01:03:35 *** WormnestAndroid has joined #openttd
02:08:41 *** D-HUND has quit IRC (Ping timeout: 480 seconds)
02:22:17 *** Wormnest has quit IRC (Quit: Leaving)
03:06:26 *** godbed has quit IRC (Ping timeout: 480 seconds)
03:41:48 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
03:42:41 *** WormnestAndroid has joined #openttd
03:46:44 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
03:47:33 *** WormnestAndroid has joined #openttd
03:51:44 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
03:52:31 *** WormnestAndroid has joined #openttd
04:21:44 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
04:24:12 *** WormnestAndroid has joined #openttd
07:03:15 *** axet has quit IRC (Remote host closed the connection)
07:13:08 *** HerzogDeXtEr has joined #openttd
07:14:54 *** Flygon has quit IRC (Read error: Connection reset by peer)
07:19:46 <petern> Oh dear, my cycling shoes are still soaking wet from yesterday...
07:19:54 <TrueBrain> you don't have an extra pair?
07:20:16 <petern> That is definitely an oversigh.
07:23:23 <TrueBrain> hmm, `IsInsideBS` is a weird function .. it depends on underflow, which is okay, but a bit nasty .. but more importantly, the second parameter is `size_t`, called `base`. The first parameter is `T` .. so those two types are not actually compatible
07:23:53 <TrueBrain> ah, it is used for integers atm ..
07:26:28 <petern> It works as long as size is not SIZE_MAX, I guess.
07:27:27 <petern> It is safe to keep an iterator for GUI use?
07:27:59 <TrueBrain> that question lacks a bit of context, I am afraid 😦
07:29:48 <petern> I'm looking at NewsItem, which is a C-style linked-list with prev/next pointers.
07:30:16 <petern> We store a pointer to the current element and can just use the prev/next pointers to move about.
07:31:04 <petern> I'm looking to C++-ify that, with a std::list, if I store a pointer to the element I'll lose the prev/next.
07:31:35 <TrueBrain> I guess an iterator can only break if the content of the list is changed without accounting for the iterator?
07:32:11 <TrueBrain> but storing an iterator in this case sounds very natural to me 🙂
07:43:14 <TrueBrain> Right, seems the ICU layouter doesn't have similar issue as Uniscribe has after all. The only way I could break it is if you do something like `abc{TINYFONT}def`. One could argue it shouldn't break between those two places .. but it does. But we don't do crap like that anyway 😛
07:43:25 <TrueBrain> good, means I don't actually have to fix anything 😛 I am good at not doing things 🙂
07:49:27 *** nielsm has quit IRC (Ping timeout: 480 seconds)
07:51:38 <andythenorth> could I just name all the wagons "wagon"?
07:51:52 <andythenorth> there are only about 30 types in Horse 😛
07:52:10 <andythenorth> and 3 colours of each type
07:55:26 <andythenorth> "MOAR as a strategy"
08:08:49 <TrueBrain> glx[d]: Was looking how to speed up mingw: `We sped up our GitHub action by adding this linker option -fuse-ld=lld` (uses the LLVM linker) .. would that be something for us?
08:40:50 <andythenorth> might pursue this idea
08:48:15 <andythenorth> wonder how this looks in JGRPP
08:51:08 <andythenorth> still clips a lot
08:51:39 <andythenorth> hmm this might work
08:51:59 <petern> Why would check why the window isn't wide enough?
08:52:36 <andythenorth> and if the screen size is too small....don't use 2x UI?
08:52:46 <andythenorth> not everything fits into everything 🙂
09:03:35 <LordAro> petern: surely there's more places that do the manual clamping before calling SetPosition?
09:04:43 <petern> Nope, most of it was already done when I made SetPosition clamp in the first place.
09:05:00 <petern> I did look and expected more!
09:08:36 <petern> Hmm potentially one in industry_gui.cpp:1892-1893
09:09:12 <petern> Very indirect though 🙂
09:09:57 <petern> That may be the wrong file name.
09:10:21 <petern> Yeah, widget.cpp:1892-1893
09:11:20 <petern> Matrix, that's newgrf station/object list. Testing...
09:13:12 <petern> No, that's not a clamp, heh.
09:13:42 <petern> I guess the comment is old, and that was already done.
09:14:08 <petern> Comment mentions GetCount() but doesn't use it.
09:26:10 *** breavyn has joined #openttd
10:14:04 *** gelignite has joined #openttd
10:43:18 *** Smedles has joined #openttd
11:33:53 <GeorgeVB> Is it possible to get Game's properties, like units (metric or imperial)?
11:34:15 <GeorgeVB> I can't find anything like that here
11:34:58 <glx[d]> That would mean desyncs
11:35:43 <petern> No. A better feature would be to allow providing a value and then using {VELOCITY} (or whatever unit string you needed) in strings. But not sure how that could be done.
11:36:54 <petern> IIRC the context here is to display an arbitrary value in the extra description text.
11:38:27 <petern> Might be a text-stack related thing, and might already be possible...
11:41:17 <glx[d]> \84 arg exist for printing speed
11:42:57 <GeorgeVB> I want to print Power and tractive effort. For some reason {POWER} does not work and TE is not supported
11:44:09 <GeorgeVB> GeorgeVB: It is related to this question
11:44:21 <glx[d]> You push the power on text stack ?
11:45:49 <GeorgeVB> Currently I form a string
11:45:49 <GeorgeVB> string(STR_PURCHASE_HINT_PROVIDES_POWER_TO, string(STR_NAME_ED2T), string(STR_UNITS_POWER_METRIC, PROP_ed2t_m_PR), string(STR_UNITS_FORCE_METRIC, round(PROP_ed2t_m_TE * KN2KGS)) and put it to the text stack
11:46:36 <GeorgeVB> STR_PURCHASE_HINT_PROVIDES_POWER_TO :{GOLD}If attached to {STRING}{}{BLACK}Power:{GOLD} {STRING}{}{BLACK}Max. Tractive Effort: {GOLD}{STRING}
11:47:02 <GeorgeVB> STR_UNITS_FORCE_METRIC :{COMMA}{NBSP}kgf
11:47:02 <GeorgeVB> STR_UNITS_POWER_METRIC :{COMMA}{NBSP}hp
11:51:27 <GeorgeVB> The reason to print power and TE rises frim the switching MU cars to wagons. But they still have power and TE, set with CB (and 0 values in props to make them wagons in depot)
11:54:08 <petern> Yeah you can't use built-in strings like that.
11:54:48 <petern> And you wouldn't need to do translation, just provide the value in the base unit and let the game convert it.
11:55:12 <GeorgeVB> How should I display Power and TE for a wagon?
11:55:49 <glx[d]> For power there's {POWER}
11:56:24 <petern> //< 9A 18: Read 2 bytes from the stack as unsigned power
11:58:55 <petern> Argh, bloody integer promotion rules
12:00:45 <GeorgeVB> nmlc ERROR: "lang\english.lng", line 1643: Provided a static argument for string command 'POWER' which is invalid
12:01:31 <GeorgeVB> This means I should push args to the stack?
12:01:43 *** breavyn has quit IRC (Remote host closed the connection)
12:02:10 <petern> Yes, you need the value to print.
12:04:09 <GeorgeVB> Ok, this may solve the power problem, but what to do with TE?
12:07:34 <petern> Looks like force is missing
12:11:42 <petern> I'm not sure how it works to go from 9A 18 to SCC_NEWGRF_PRINT_WORD_POWER.
12:29:53 <petern> GeorgeVB: , if you fancy coding it as "9A 21" and sending me a test NewGRF, I can check if this works 🙂
12:31:23 <petern> Also I assumed WORD is enough for force...
12:53:27 <glx[d]> included a standalone windows build for easier testing
13:19:24 <petern> Hmm, of course, uint16 + uint16 = int.
13:19:41 <petern> Now, why can't I get my compilers to complain about the same things...
13:20:16 *** gelignite has quit IRC (Quit: Stay safe!)
13:43:03 *** WormnestAndroid has quit IRC (Remote host closed the connection)
13:43:08 *** WormnestAndroid has joined #openttd
13:52:03 <petern> -Wsign-conversion didn't do much.
13:52:51 <petern> Ah, it deletes my flags. Hmm.
13:54:47 <petern> How do I add extra flags? 😮
13:54:49 <glx[d]> if you want to change flags it's in D:\developpement\GitHub\glx22\OpenTTD\cmake\CompileFlags.cmake
13:57:27 <petern> `string_func.h:86:11: warning: implicit conversion changes signedness: 'long' to 'size_t' (aka 'unsigned long') [-Wsign-conversion]
13:58:23 <petern> So const char * - const char * is a long not a size_t.
13:58:37 <petern> I guess it'll go negative if it's the other way around.
13:59:07 <TrueBrain> should be a `ptrdiff_t` not?
14:02:28 <glx[d]> ok let's do a `pacman -Syuu` in my mingw console
14:02:37 <petern> Well, -Wsign-conversion is not usable 🙂
14:05:34 <petern> Oops, the problems list has crashed VS Code.
14:05:54 <andythenorth> expand on select?
14:06:02 <glx[d]> I want to try some things, like using ninja instead of make, and also the lld thing
14:07:12 <petern> Variable-height options causes fun in our UI code 🙂
14:07:35 <petern> WWT_MATRIX won't handle it for sure.
14:08:03 <glx[d]> the screenshot seems to show it handles it
14:08:22 <glx[d]> can't tell about the scrollbar
14:08:25 <petern> andythenorth is quite good at mockups 🙂
14:08:45 <andythenorth> I have not extended a grf string suffix for engines
14:09:08 <andythenorth> I am trying to find some way to put extra info next to the sprite + name
14:09:15 <andythenorth> but keeping the buy menu simple
14:09:25 <petern> I mean, it looks quite nice but if you have to click it to show it, it might as well just appear in the details pane.
14:09:45 <andythenorth> I would just do that...but cargos 😛
14:10:04 <andythenorth> this list though
14:10:11 <petern> Can we change that cargo list. It's unreasonable.
14:10:26 <andythenorth> what triggers this handling?
14:10:49 <petern> Also it's meant to do that, yes.
14:11:13 <andythenorth> is it just pax, mail that triggers the 'all but'?
14:11:18 <petern> if (CountBits(cmask ^ lmask) < CountBits(cmask) && CountBits(cmask ^ lmask) <= 7) {
14:11:24 <petern> I wonder what the <= 7 is for.
14:11:27 <andythenorth> hmm shame we only have {TINYFONT} and not {SMALLFONT}
14:11:34 <andythenorth> do I have to give up my pixel font? 😦
14:11:44 <glx[d]> oh maybe the 7 is from before 64 cargo types
14:11:58 <andythenorth> so double it to 14?
14:12:06 <glx[d]> what was the limit before ?
14:12:17 <petern> I don't think there was a limit.
14:12:47 <andythenorth> cargos used to be 32
14:12:59 <petern> Yes but even then 32-7 is 25.
14:13:18 <andythenorth> `Refittable to: Many things`
14:13:45 <petern> Unfortunately the commit doesn't way why it does it.
14:13:46 <TallTyler> We have the cargo filter now, could possibly drop the text entirely
14:13:47 <andythenorth> hmm do we have a string code for a horizontal rule?
14:14:07 <TallTyler> I always use the cargo filter when buying vehicles
14:14:14 <glx[d]> there was 12 or 32 at that time
14:15:12 <petern> It was 32 back in 2010.
14:16:23 <glx[d]> it should just compare the number of refittable to vs non refittable and use the shortest list
14:17:18 <glx[d]> no need to test for <= 7
14:21:49 <petern> That's what it did before the 7 limit was added.
14:23:00 <petern> I suspect it was chosen because it's possibly better to display 7 includes than 5 excludes.
14:23:36 <petern> If anything it should compare the delta between include & exclude, and go with include if the delta is <= 2 or something small.
14:24:01 <petern> Heh, of course it already compares, so..
14:24:45 <petern> Something like `if (CountBits(cmask ^ lmask) + 2 < CountBits(cmask)) {`
14:25:29 <petern> andythenorth: is that steeltown?
14:26:21 <andythenorth> steeltown 5, unreleased
14:30:11 <Rubidium_> petern: storing the iterator of std::list is fine, as long as you can guarantee the item it is pointing to is not removed; or in spec speak: "References and iterators to the erased elements are invalidated. Other references and iterators are not affected."
14:33:12 <petern> Also is Horse 4 not on bananas...?
14:39:07 <TrueBrain> I would almost say it should be the inverse: if there are more than 3 or 4 "but ..", just list them all
14:39:14 <TrueBrain> these "all but <long list>" are so silly 😛
14:39:32 <andythenorth> it's silly vs. silly
14:39:58 <TrueBrain> yeah, but "all but" requires actual reading
14:40:06 <TrueBrain> I almost mistook that list for things you could refit too
14:40:29 <TrueBrain> I guess making the "but" another colour also wouldn't hurt 😛
14:40:41 <andythenorth> how about grey? 😛
14:41:06 <TrueBrain> visually, a very difficult line, that "refittable" 😛
14:42:21 <petern> Yes, but better than in andy's screenshot.
14:42:30 <andythenorth> it was ok when vehicles had one cargo 😛
14:44:11 <petern> Well, not really better.
14:44:17 <petern> All but a long list is pretty useless less.
14:44:36 <petern> List all cargo types with a tick next to them.
14:47:48 <andythenorth> hmm inspiration from JGRPP?
14:54:19 <glx[d]> glx[d]: and the upgrade is still not finished
14:55:08 <glx[d]> almost 1h and "( 95/225) upgrading..."
14:55:48 <glx[d]> mingw is slow for everything 🙂
15:00:08 <JGR> Are you using mingw through cygwin or msys or something like that?
15:00:25 <glx[d]> it's the mingw console
15:00:33 <JGR> When I run mingw on Linux it doesn't seem that slow at all
15:00:49 <glx[d]> mingw and disk access are super slow on windows
15:01:03 <glx[d]> because posix layer on top of windows API
15:01:36 <JGR> It ought run at a reasonable speed through WSL though?
15:02:47 <glx[d]> wsl also has disk access issues if using windows file system, that's why MSVC first rsync the files inside wsl filesystem
15:22:32 <TrueBrain> MSYS2 fixed a lot of performance issues, but .. yeah, it is kinda obsolete now with WSL
15:25:16 <petern> Expecting a load of signed/unsigned errors :p
15:51:16 <glx[d]> ah seems I have GCC 13.1.0
15:51:26 <TrueBrain> Wtf is freight grey...
15:52:29 <TallTyler> What about Ctrl+click on a parent to buy a random subvariant?
15:52:41 <TallTyler> Ctrl+click on the buy button that is
15:52:52 <glx[d]> you want a fight with authors ?
15:53:14 <andythenorth> I was big on that idea
15:53:27 <andythenorth> grf could have a callback to provide the list of candidates
15:54:09 <JGR> It seems a lot of work for something which users will likely rarely use
15:55:18 <petern> Oh, the assert itself is wrong 😄
15:55:18 <andythenorth> TrueBrain: it's not company colour grey, it's grey range I made 😛
15:55:23 <andythenorth> CC grey is weird
15:55:24 <glx[d]> `[ 8%] Generating orig_dos.obs base[s e t8 %m]e taCdoaptyai nfgi lseta
15:55:24 <glx[d]> tionlist/info.nut regression file` <-- using -j6 is fun
15:56:07 <andythenorth> I still think that random wagons are a bad idea and I _should_ probably delete them
15:56:16 <TrueBrain> andythenorth: But it is not like the user actually cares there is another grey :p
15:56:17 <andythenorth> anything that can't be explained to the player is bad in a game
15:58:53 <petern> Ah, SetPosition needs to accept int to allow the clamping to work.
15:59:21 <glx[d]> oh I have warnings with mingw
16:00:20 <andythenorth> this does look nice though?
16:00:31 <andythenorth> yeah but it's confusing
16:00:41 <andythenorth> when you build a wagon and they're different colours
16:00:51 <andythenorth> looks like a bug?
16:00:54 <glx[d]> 3 redondant move in return statement
16:01:08 <petern> TBH, it's something I'd expect from a "realism" set, not IH 🙂
16:01:19 <petern> But OTOH, it is a nice bit of simple variety.
16:01:28 <andythenorth> IH has a carefully hidden core of realism :
16:01:35 <andythenorth> just alternate realism
16:01:48 <glx[d]> I think it should be the only option 😉
16:02:04 <andythenorth> it would be ok if it was the only option
16:02:50 <andythenorth> but I also want to be able to do this 😛
16:04:17 <andythenorth> then it's confusing in auto-replace
16:05:03 <petern> Yeah, why have it manually selected?
16:05:05 <TallTyler> Autoreplace is probably the best argument for one vehicle with random sprites
16:06:09 <TrueBrain> You should use it to age wagons
16:06:19 <TrueBrain> The older, the more grey
16:06:20 <andythenorth> "your train got rusty"
16:06:25 <andythenorth> then we could have "repaint"
16:06:30 *** WormnestAndroid has quit IRC (Remote host closed the connection)
16:06:32 *** WormnestAndroid has joined #openttd
16:06:36 <andythenorth> as a way to solve "too much money"
16:06:37 <glx[d]> and cleanup on service
16:09:02 <andythenorth> could just give them arbitrary different names
16:09:13 <andythenorth> so the CC is called one thing and the brown is called another
16:09:26 <andythenorth> Horse already does that for some wagons where the colour changes
16:12:02 <TallTyler> Oops, commented at the same time
16:15:40 <andythenorth> meh, if I could just improve auto-replace explanation, I think these would be ok
16:15:52 <andythenorth> can always do some parameter to turn the whole thing off, yes/no? 😛
16:19:14 <andythenorth> and I will put a "Livery:" string in the extra text
16:20:58 <ag> I wish openttd had visual effects showing aging of vehicles
16:21:16 <ag> Which would require a lot of work for nothing
16:22:16 <ag> Joker Patchpack used to have visual effect showing unused tracks with a different colour or something
16:26:06 <andythenorth> I can detect the current company colour and name it
16:26:44 <andythenorth> not sure what to call the brown / grey / black
16:26:51 <ag> I think you are overcomplicating it
16:27:18 <andythenorth> I have only spent 5 days on how to name this in the buy menu so far 😛
16:27:56 <ag> I just want rusty looking bogies when I use the same rolling stock for more than 25 years
16:29:58 <petern> My other argument against changing it 😉
16:32:25 <TallTyler> A classic argument. 🙂 Post that in the PR!
16:32:35 *** Smedles has joined #openttd
16:32:35 <glx[d]> linking with mingw is so slow
16:32:57 <glx[d]> I think it's in link phase for more than 30m now
16:33:31 <petern> I think he just compared it with the "top-level" construction windows, not things with extra choices.
16:33:36 <petern> Now, our Signal UI, on the other hand...
16:34:22 <glx[d]> let's try other options
16:35:19 <TallTyler> What did the signal UI look like in TTD?
16:36:03 <TallTyler> Oh, only one signal type?
16:36:04 <LordAro> TT didn't even have one-way signals
16:36:59 <glx[d]> TTD introduced one-way, but AI still used the same layouts (just with one-way)
16:46:36 <petern> max / min, what's the difference...
16:48:15 <andythenorth> so do we know the x dimension of strings when drawn?
16:55:45 <petern> It's stupid how confusing min/max is, because it should be obvious every time...
16:58:35 <TrueBrain> last time I mentioned that, I got laughed at ... was not a week ago 😛
16:58:43 <TrueBrain> the tooltip in VSCode is also insulting
16:59:20 <petern> > This does what you think it does.
16:59:49 <petern> The whole documentation there is a bit... slack?
17:00:04 <petern> > __a – A thing of arbitrary type.
17:00:04 <petern> > __b – Another thing of arbitrary type.
17:00:22 <TrueBrain> feels like someone had a friday commit
17:00:25 <TrueBrain> and someone else approved it
17:00:30 <TrueBrain> and he was like: REALLY?! Well, okay ...
17:01:15 <glx[d]> VS just shows the signature
17:01:20 *** Wormnest has joined #openttd
17:03:13 <andythenorth> colour scheme? livery? colourway? theme?
17:03:13 <petern> Hopefully it compiles without any signed/unsigned warnings :p
17:03:19 <andythenorth> what to call the colour sets?
17:28:18 <pickpacket> is the game making a bunch of calculations at the end of the year for the summary?
17:28:37 <pickpacket> or are those calculations done all the time?
17:28:45 <TrueBrain> Rubidium: you noticed you don't actually use the `extra-cmake-parameters` you added, right? 🙂
17:29:11 <TrueBrain> and the 2019s are not 2019 😛
17:29:51 <TrueBrain> you now just defined a variable in the matrix; you also have to actually use it when running cmake to make it effective 🙂
17:30:44 <Rubidium_> I hope this is better
17:31:05 <TrueBrain> at least an higher chance of doing what you want 🙂
17:32:24 <TrueBrain> owh, I see why you did it
17:32:32 <TrueBrain> euh .. we can fix that .. let me think a sec ..
17:33:04 <petern> Disabling precompiled headers for windows?
17:33:28 <andythenorth> what to name this type of colour variant?
17:34:05 <TrueBrain> sorry Rubidium, I could have read more carefully before replying 🙂
17:54:52 <glx[d]> hmm the slowest part of linking with mingw doesn't seem to be ld, but ar
17:58:52 <LordAro> ar is a very basic archiver
17:58:57 <LordAro> must just be filesystem IO speed
18:04:44 <andythenorth> linking on macos is horribly slow
18:04:52 <andythenorth> I imagine that's unrelated
18:07:27 <TrueBrain> glx[d]: maybe we should make a ramfs for mingw to compile in 😛
18:07:31 <TrueBrain> no clue if that is possible with windows 😄
18:08:24 <TrueBrain> I guess building a release might also be quicker in that case
18:08:30 <TrueBrain> as debug builds ofc are HUGE
18:39:22 <TrueBrain> lol, a blog post about making smaller C++ binaries .. `Use -m32 to compile a 32-bit binary. 32-bit binaries are smaller than 64-bit binaries because pointers are shorter. `
18:39:32 <TrueBrain> somehow, I think that is cute 🙂
18:39:35 <DorpsGek> - Update: Translations from eints (by translators)
18:59:18 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
19:09:38 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
19:11:19 *** WormnestAndroid has joined #openttd
19:26:52 *** Wormnest has joined #openttd
19:28:23 *** gelignite has joined #openttd
19:36:10 <glx[d]> TrueBrain: not sure, release builds use LTO
19:36:52 <glx[d]> but I just started a local test with lld and release
19:39:48 <glx[d]> the weird thing is I had a ~15m gain when trying just with ldd (from 39m1 to 24m50)
19:41:07 <glx[d]> but #10782 is totally untested 🙂
19:41:39 *** axet has quit IRC (Quit: Leaving.)
19:42:35 <glx[d]> so nice I get even more warnings with release build
19:55:04 <glx[d]> ```ld.lld: error: undefined symbol: WinMain
19:55:04 <glx[d]> >>> referenced by C:/M/B/src/mingw-w64/mingw-w64-crt/crt/crtexewin.c:70
19:55:04 <glx[d]> >>> libmingw32.a(lib64_libmingw32_a-crtexewin.o):(.text.startup)
19:55:04 <glx[d]> collect2.exe: error: ld returned 1 exit status
20:03:46 <petern> That wiki editor is a bit painful 😦
20:07:21 <TrueBrain> TallTyler: Well, no, not what I have been asking for really 😄 I am curious if the overall text is okay, if we are missing anything, etc :p
20:09:14 <TrueBrain> Hihi, tnx frosch , was wondering how long it would take you to tell me what flag is useful :p :p
20:09:46 <TrueBrain> Is invalid not interesting btw?
20:10:47 <frosch> if you would upload crash-reports you could include the gamelog, but the grf-flags are not interesting
20:11:37 <TrueBrain> And it sends a survey upon crash, but it doesn't send the gamelog or backtrace atm
20:12:04 <frosch> pretty sure the gamelog contains paths
20:12:22 <TrueBrain> Yeah, it has a high risk of doing so 🙂
20:12:34 <TrueBrain> Took effort to ensure other parts didn't :p
20:13:34 <frosch> i wonder why noone from the c++17 refactor gang removed GCF_COPY yet :p
20:14:51 <frosch> hmm, maybe it is doing something else than i remembered
20:15:21 <petern> Looks like it is more than just "do we need to free this data"!
20:15:41 <petern> Also `c->info = f->name` looks wrong.
20:16:48 <Rubidium_> frosch: I thought about it, didn't immediately understand what was going on there and just continued with something else
20:17:09 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
20:17:45 <Kuhnovic> The GUI redesign rabbit hole is deeeeeep
20:17:54 <frosch> TrueBrain: how did you end up using ruby?
20:18:18 <TrueBrain> Really didn't want to deal with CSS
20:18:41 <frosch> ah, it's that jekyll thingie again, so only build time
20:19:04 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
20:19:09 <petern> Kuhnovic: and it doesn't look better IMHO.
20:19:48 <Kuhnovic> Thanks haha 😛 I'm still experimenting, not even sure if I want to continue with this at all
20:32:42 <petern> ^ Good example of what not to write.
20:37:31 <andythenorth> when everything is written by GPT, how will GPT train itself?
20:42:45 *** nielsm has quit IRC (Ping timeout: 480 seconds)
20:49:38 <Eddi|zuHause> it will be a self-fulfilling prophecy
20:50:43 <andythenorth> well surely it can train itself on future output in that case?
20:50:51 <andythenorth> so it can know what it will write in future?
20:57:17 *** Wormnest has joined #openttd
21:16:00 <andythenorth> pff Horse now 22.7 MB
21:16:04 <andythenorth> this is getting silly
21:16:23 <andythenorth> just to add buy menu additional text to every wagon
21:17:06 *** WormnestAndroid has joined #openttd
21:19:02 <petern> How many engines, 10,000?
21:20:02 <andythenorth> nmlc info: Train items: 2666/65420
21:31:47 *** keikoz has quit IRC (Ping timeout: 480 seconds)
22:03:02 *** ChanServ sets mode: +v tokai
22:08:01 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
22:09:39 *** tokai|noir has quit IRC (Ping timeout: 480 seconds)
22:25:57 *** gelignite has quit IRC (Quit: Stay safe!)
22:53:49 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
22:54:27 <andythenorth> you drew things? 🙂
22:55:08 <andythenorth> pff how do I read just one string off the text stack? 😛
22:55:25 <andythenorth> store(string(FOO) << 16, 256)
22:58:09 <petern> If by "draw" you mean "told OpenTTD to draw", I guess so?
22:59:06 <Eddi|zuHause> i think you might need to store FOO, not string(FOO)
22:59:27 <Eddi|zuHause> not really sure though
22:59:34 <Eddi|zuHause> haven't done this in ages
23:00:30 <andythenorth> ` STORE_TEMP(string(STR_POWER_SOURCE_${power_source_data[0]})
23:00:30 <andythenorth> | (${power_source_data[1]} << 16),
23:00:30 <andythenorth> ${'0x10' + str(repeat.power_source_data.index)}),
23:00:42 <andythenorth> but in that case I'm pushing 2 things to the register
23:01:08 <andythenorth> in the case of pushing one it's puzzling
23:01:08 <Eddi|zuHause> the first string would be the << 0, not the << 16
23:10:57 <Eddi|zuHause> the newgrf access to the "text stack" is probably the weirdest bit of exposed implementation detail
23:11:31 <andythenorth> for my next case I need to put a string on the stack, then have that string pull things from the stack
23:14:21 <glx[d]> a syntax like GSText would be easier to use
23:16:33 <andythenorth> it's always stuck out as odd 🙂
23:16:41 <andythenorth> we could wrap it up in some nml sugar
23:16:59 <andythenorth> having to know about sizes and shifts is very...not like nml
23:17:30 <glx[d]> yeah like string(ID_OF_A_STRING_NEEDING_PARAMS, the_param)
23:20:29 <Eddi|zuHause> that exists, but only for static params
23:20:32 <glx[d]> or at least some SetTextParam(0, blah)
23:21:02 <glx[d]> which handles the register count and shifting
23:21:15 <Eddi|zuHause> it gets partly complicated because you have to know the size of the value being pushed
23:22:08 *** D-HUND is now known as debdog
23:22:28 <andythenorth> it's quite faffy to template
23:22:48 <andythenorth> if it's a list of words, it has to alternate << 0 and << 16
23:22:59 <andythenorth> and if it's a list of unequal sized things...good luck 😛
23:23:25 <Eddi|zuHause> andythenorth: in general, if all the values are 16 bit, the "text stack" looks like this: [STORE_TEMP(A | (B<<16), reg_1), STORE_TEMP(C | (D<<16), reg_2), STORE_TEMP(E | (F<<16), reg_3)]
23:23:53 <andythenorth> that's what the code I pasted above is doing
23:24:10 <Eddi|zuHause> so if C is a substring that consumes a value, D is that value being consumed
23:24:19 <andythenorth> it's got a python tuple in the case above
23:24:39 <Eddi|zuHause> and the main string continues with E
23:26:25 <Eddi|zuHause> and the real baad part comes wthen C is only sometimes a string that consumes a value
23:26:46 <Eddi|zuHause> then you need to put E in different positions depending on the value of C
23:27:10 <Eddi|zuHause> i think we removed skip
23:27:48 <petern> Hmm, is there a function to strip bad characters from NewGRF strings?
23:27:53 <glx[d]> skip is still there for newgrf
23:27:55 <andythenorth> pff is the stack read high to low or low to high?
23:27:57 <petern> Basically anything like a control code.
23:28:48 <Eddi|zuHause> andythenorth: the point of a "stack" is to read the values in the opposite order than you pushed them
23:29:39 <glx[d]> and text stack is not a stack, it's an array 😉
23:29:56 <Eddi|zuHause> that would've been my next sentence
23:30:00 <andythenorth> too much bedtime 😛
23:30:14 <Eddi|zuHause> the question is, why do you need to know?
23:30:36 <andythenorth> getting nonsense results
23:30:55 <andythenorth> but I am trying to do something that is probably unwise
23:31:08 <andythenorth> I have parameterised string as a parameter to another string
23:31:13 <Eddi|zuHause> those are usually solved by stepping backwards and looking at the logic from a higher standpoint
23:32:01 <Eddi|zuHause> could be lots of things, like wrong order, misplaced operator or brackets
23:33:14 <andythenorth> can be tomorrow's problem
23:33:18 <andythenorth> well, later today's
23:41:14 <glx[d]> ld and release builds fine```real 21m47.331s
23:41:14 <glx[d]> ``` lld and release fails to link
23:57:21 <petern> Very fancy name for a signal-waypoint.
continue to next day ⏵
