IRC logs for #openttd on OFTC at 2019-03-02
⏴ go to previous day
00:00:32 <TrueBrain> LordAro: yeah, so it was from back in the days we shipped SDL.dll next to openttd.exe
00:00:39 <TrueBrain> Gwyd: that's ... what he said?
00:01:18 <TrueBrain> LordAro: so .. burn it with fire?
00:01:27 <peter1138> Gwyd came to the right place, nice.
00:01:47 <Gwyd> great I'm here so I read that pull request discussion and the "train state" thing seems like a more useful thing than being able to query/compare speed limits
00:02:43 <Gwyd> i.e. asking if the train is accelerating, braking, trying it's best to stay at a speed or stopped (for different reasons)
00:03:31 <Eddi|zuHause> Gwyd: yeah, it'll be a bit tricky extracting that information from the code, though
00:03:37 <TrueBrain> glx: seems like the same "bug" happens for freetype. It feels like pkg-config is going wrong here on OSX
00:04:01 <Samu> i can finally name my groups!
00:04:17 <Gwyd> Eddi: yeah, because a train may be slowing down, but only due to a hill or something
00:04:18 <Samu> without worrying about it being unique
00:04:30 <TrueBrain> glx: tomorrow I will fix the CI for Linux, so you can peak there too what happens :)
00:04:54 <peter1138> Samu, needs someone to review it.
00:04:56 <glx> well pkgconfig does the same for mingw, but at least all libs are installed in the same location
00:04:58 <peter1138> Alas, nobody cares ;)
00:05:25 <TrueBrain> glx: yeah .. possibly pkg-config for lzo is just broken or something
00:05:32 <TrueBrain> not sure .. needs more debugging :)
00:05:53 <Samu> btw Eddi|zuHause The cheated value is 21, not 0
00:05:55 <glx> for vcpkg detection use find method
00:06:09 <TrueBrain> vcpkg comes with CMake for every package, basically
00:06:12 <TrueBrain> but vcpkg patches it
00:06:13 <peter1138> Okay so the issue with saving the deque content is we have to know, in advance, at compile time, what the data type is.
00:06:15 <TrueBrain> so that is not a real surprise
00:06:19 <LordAro> TrueBrain: currently working out if i can get rid of sdl.cpp/h entirely
00:06:31 <TrueBrain> anyway, time to get some sleep
00:06:36 <peter1138> This is not possible with out sl_desc arrays.
00:06:46 <glx> I meant for VS builds using vcpkg, our script use find method to detect lzo :)
00:06:51 <Samu> someone fix that table :O
00:07:03 <TrueBrain> glx: yes; vcpkg adds cmake files for every package you install
00:07:07 <Eddi|zuHause> peter1138: not sure i understand the problem
00:07:15 <TrueBrain> so every library installed with vcpkg, will be found by find()
00:07:28 <TrueBrain> but vcpkg patches that, if the library doesn't support it
00:07:49 <TrueBrain> glx: owh, you meant the other find()
00:07:53 <TrueBrain> sorry, too many finds in this world
00:08:02 <peter1138> Eddi|zuHause, because the type of the std::deque is a template, it has to be there at compile time.
00:08:03 <TrueBrain> as it should be a cmake :P
00:08:12 <TrueBrain> owh, findPackage() ofc
00:08:16 <TrueBrain> strange that is not working
00:08:30 <peter1138> But saveload stuff is detached and hidden behind layers.
00:08:35 <Eddi|zuHause> peter1138: so the saveload code must know the exact instantiation of the template?
00:08:37 <TrueBrain> anyway, really off now
00:08:49 <glx> -- LZO_INCLUDE_DIRS: D:/developpement/GitHub/vcpkg/installed/x64-windows-static/include
00:09:16 <Eddi|zuHause> so it must be put into some header somewhere
00:09:17 <peter1138> Eddi|zuHause, the only way is to teach the saveload code about every possible type that we want to saveload.
00:09:24 <peter1138> Some header, yeah, not quite.
00:11:07 <peter1138> Eddi|zuHause, each of those SlDequeHelper lines has a specific type.
00:11:50 <peter1138> Hmm, even then I don't think it would work.
00:12:23 <peter1138> Ah, SlSaveLoadConv() would have to know about the datatype.
00:12:48 <peter1138> Although that way it would only work up to uint64.
00:12:57 <peter1138> This saveload system is a bit ... baroque ;)
00:13:05 <Eddi|zuHause> so you need a SL_VAR_SOMESTRUCT and instantiate SlDequeHelper<SomeStruct>?
00:13:53 <peter1138> actually it would just be a custom SlDequeHelper than knows the contents the struct.
00:14:35 <peter1138> So... easier to have either 2 deques, or bitstuff it.
00:15:00 <Eddi|zuHause> i think we need to zoom out one or two levels before seeing how to properly fit this in there
00:15:20 <peter1138> Well there was an alternative :/
00:15:40 <peter1138> Take it out of the sl_desc array and manually handle i.
00:34:17 <peter1138> ^^ two separate deques.
00:35:36 <peter1138> ib4 use static arrays
00:35:42 <peter1138> Although actually...
00:38:40 <Eddi|zuHause> i'm really unsure about either, but the bitstuffing one seems less future proof
00:40:27 <peter1138> Hmm, I could use a union instead ;)
00:41:26 <LordAro> hmm, mouse is very flickery
00:41:31 <LordAro> built & linked though
00:43:23 <Eddi|zuHause> intuitively i think the SLE_DEQUE() stuff needs a nested struct decomposing it into SLE_UINT8/32
00:44:33 <peter1138> Yeah but that's just not possible with how that works.
00:45:09 <peter1138> Hmm, maybe a 3rd way :-)
00:46:37 <glx> LordAro: by default it's a debug buid, unless you configured with -DCMAKE_BUILD_TYPE=Release
00:46:57 <Eddi|zuHause> the whole saveload code screams "REWRITE ME WITH TEMPLATES"
00:47:29 <LordAro> glx: so the flickery seems to be the win32 video driver, nothing i did
00:47:40 <LordAro> but i definitely don't have debugging symbols
00:48:55 *** drac_boy has joined #openttd
00:50:27 <drac_boy> not sure if its a localized word or not so much but anyway .. are tipper wagons basically flatcars-top or they're rather more dedicated (alike to tankers with minimal/no chassis)
00:50:55 <glx> and it's a dynamic build too, but that's ok for a local build
00:54:11 *** supermop_work_ has quit IRC
01:04:13 *** Thedarkb-T60 has joined #openttd
01:06:20 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
01:19:59 <peter1138> Hmm, this isn't going so well
01:29:43 <glx> btw LordAro, I have some PRs on TB fork
01:39:07 <drac_boy> samu I hope you're not in dreamland now :)
01:42:23 <Samu> it now uses OTTDStringSorter inside win32.cpp
01:47:59 <Samu> Function Name Total CPU [unit, %] Self CPU [unit, %] Module
01:47:59 <Samu> | - std::_Deque_alloc<std::_Deque_base_types<char const *,std::allocator<char const *> > >::_Get_data 39 (6,55%) 37 (6,22%)
01:48:16 <Samu> it's still Deque_alloc who's taking most time
01:48:19 <DorpsGek_II> [OpenTTD/OpenTTD] nikolas updated pull request #7086: Change #6173: Update SDL driver to use SDL 2.0 https://git.io/fhamZ
01:50:30 <DorpsGek_II> [OpenTTD/OpenTTD] nikolas updated pull request #7086: Change #6173: Update SDL driver to use SDL 2.0 https://git.io/fhamZ
01:51:35 <peter1138> in src/strings.cpp:794
01:51:50 <peter1138> std::stack<const char *> str_stack;
01:52:00 <peter1138> std::stack<const char *, std::vector<const char *>> str_stack;
01:52:12 <peter1138> See if it makes a difference.
02:01:41 <peter1138> Probably makes no difference.
02:04:39 <Samu> too fast, which is weird
02:05:59 <Samu> Function Name Total CPU [unit, %] Self CPU [unit, %] Module
02:06:00 <Samu> | - OTTDStringCompare 513 (85,93%) 95 (15,91%)
02:07:06 <peter1138> It is quite a bit faster.
02:07:22 <peter1138> Not as fast as comparing directly without formatting, but still.
02:07:48 <Samu> Function Name Total CPU [unit, %] Self CPU [unit, %] Module
02:07:48 <Samu> | - std::_Vector_alloc<std::_Vec_base_types<char const *,std::allocator<char const *> > >::_Get_data 10 (7,25%) 9 (6,52%)
02:08:01 <Samu> this is for FilterSignList()
02:08:25 <peter1138> %ages are not really relevan.t
02:09:06 <Samu> let me try now release build without visual studio attached
02:09:14 <peter1138> If performance is improved in one area, percentage-wise it will increase elsewhere, but it hasn't actually increased in real tersm.
02:09:54 <peter1138> I'm see 66/100ms per sort instead of 350ms
02:10:34 <Samu> opening the sign window list with 62k signs, had a little stall of 62 ms
02:11:24 <Samu> because it sorts afterwards
02:13:45 <Samu> that is slightly better than
02:14:11 <peter1138> Hmm, greater difference for me.
02:15:57 <Samu> 465 ms, it's speeding up, the AI is clearing signs atm
02:21:20 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN opened pull request #7305: Codechange: Make std::stack use std::vector container in string formatting/drawing. https://git.io/fhAVC
02:24:37 <Samu> no idea what's a vector not a stack or a deque, but cool stuff
02:25:15 <peter1138> STL stuff. Different implementations of similar things.
02:25:53 <peter1138> deque is better for larger structures.
02:29:16 <Samu> what do squirrel lists use?
02:32:09 <peter1138> A std::map stores a relationship from one type to another.
02:32:24 <peter1138> Hmm, yeah, scriptlist uses both.
02:33:43 <Samu> { // ordered red-black tree of {key, mapped} values, unique keys
02:34:08 <Samu> { // ordered red-black tree of key values, unique keys
02:35:04 *** snail_UES_ has joined #openttd
02:35:32 *** Wormnest has joined #openttd
02:35:59 <LordAro> optimising for the sake of optimising is often a waste of time
02:36:09 <LordAro> especially when you're not actually profiling properly
02:36:20 <LordAro> Samu: look up "premature optimisation"
02:36:49 <DorpsGek_II> [OpenTTD/OpenTTD] nikolas updated pull request #7086: Change #6173: Update SDL driver to use SDL 2.0 https://git.io/fhamZ
02:51:38 <Samu> so basically someone said one thing and everyone distorted it
02:52:01 <LordAro> i mean, that article is precisely the opposite of the point i was trying to get across to you
02:52:09 <glx> pff I was searching a setting in the GUI, but it's a console/config_file only for 5 years
02:52:14 <LordAro> but since you didn't bother reading it anyway, i'm not sure how much i care
02:52:18 *** Supercheese has joined #openttd
02:52:46 <peter1138> I don't understand why we have a filterable settings GUI... and then removed a load of settings from it :/
02:53:19 <glx> yes it was an expert one, order.selectgoods
02:53:45 <Samu> i like that one, I occasionaly use it
02:54:00 <glx> yes it's a nice one, but it's hidden
02:55:12 <Eddi|zuHause> peter1138: because SETTINGS ARE EVIL!!!
02:55:30 <glx> improved loading is hidden too
02:55:33 <Eddi|zuHause> people can't be trusted with settings
02:55:47 <Samu> make settings level, similar to debug level
02:56:03 <Samu> 1 - basic settings, 10 - everything
02:56:04 <peter1138> We have. Basic. Advanced. Expert.
02:56:17 <peter1138> Many settings are in the wrong category.
02:56:28 <Eddi|zuHause> all people are experts, anyway
02:58:19 <Eddi|zuHause> ok, i stop staring at the saveload code now.
02:58:55 <glx> all removed were expert, except selectgoods which had no category (so basic I guess) and has been moved to expert during the removing
02:59:39 <Eddi|zuHause> i can't remember which side of that discussion i was on
03:00:25 <Eddi|zuHause> i almost definitely wasn't involved in that
03:09:15 <Samu> can i suggest inflation off as default?
03:09:28 *** Thedarkb-X40 has joined #openttd
03:11:42 <Samu> a setting that could be removed from the gui, imo
03:11:54 <Samu> "Disable electric rails"
03:12:28 <Samu> i've almost never used it
03:12:43 <peter1138> It's a backwards compatibility thing.
03:13:15 *** Thedarkb1-X40 has joined #openttd
03:15:03 <peter1138> ^^ that is pretty ugly too :(
03:16:34 <Eddi|zuHause> "RealSave_VEHS" uhm, what? :p
03:17:41 <peter1138> Borrowed it from station_sl.cpp
03:20:10 <Eddi|zuHause> "Don't put a person in a position where they have only bad decisions to choose from"
03:21:10 <Samu> why is 7305 separate from 7304?
03:21:34 <peter1138> Two separate changes.
03:22:14 <peter1138> If 7304 is in place, 7305 has almost zero impact on your sorting.
03:22:50 <peter1138> (But other sorted things might benefit, ever so slightly...)
03:22:53 <Samu> didn't test the color thing
03:23:47 <peter1138> If 7304 is applied, then the changes in 7305 will NOT affect anything, because 7304 bypasses is that all anyway.
03:26:00 <Samu> testing 7305 separately, brb
03:27:50 <Samu> better than 3378 i reported initially
03:29:32 <Samu> 7304 got those results I posted
03:29:43 <Samu> but will test again, just to confirm
03:29:47 <Eddi|zuHause> youtube suggestions horrible as usual
03:38:59 <Samu> why is it faster? no idea
03:40:55 <Samu> 2.9 frames per sec, still unbearable
03:41:28 <peter1138> Sure, but there'll be an improvement at less extreme cases too.
03:46:52 *** Beerbelott has left #openttd
04:02:07 <Samu> gonna test Eddi|zuHause bfsriver stuff
04:06:11 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN updated pull request #7304: Codechange: No need to buffer and format sign strings, they are C-style strings to start with https://git.io/fhAE9
04:10:45 <Samu> is it just me or it hangs?
04:11:43 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN updated pull request #7304: Codechange: No need to buffer and format sign strings, they are C-style strings to start with https://git.io/fhAE9
04:15:47 <Samu> gonna wait at most 10 minutes
04:16:06 <Samu> seems to hang on generating rivers for 4096x4096
04:19:30 <Samu> these rivers look nice, but they're quite short
04:19:50 <Samu> and they spring in the open, not near any mountains
04:20:16 <Samu> they're also not lock-ready
04:26:26 <peter1138> That isn't a thing.
04:29:38 <Samu> doesn't like mountains much
04:29:45 <Samu> usually on the low ground
04:30:37 <Samu> smooth terrain helps a lot, I see
04:32:10 <Samu> very smooth is perfect, even better
04:32:27 <Samu> goes up mountains nicely
04:33:04 <Samu> lets try a map heeight of 255
04:36:53 <Samu> but rivers sometimes end in weird spots
04:37:02 <Samu> or are sized 3 tiles or so
04:41:59 <Samu> it's impressive, but it makes many veins in a tight space and then it makes nearly nothing in a large block
04:43:23 <Samu> but terrain needs to be smooth pretty much
04:43:33 <Samu> correction, "very smooth"
04:43:58 <Samu> better than my lock-ready enforcements
04:52:13 *** tokai|noir has joined #openttd
04:52:13 *** ChanServ sets mode: +v tokai|noir
04:55:45 <Samu> weird, i get more rivers on mountain terrain than hilly
05:23:17 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN updated pull request #7304: Codechange: No need to buffer and format sign strings, they are C-style strings to start with https://git.io/fhAE9
05:34:01 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on pull request #7304: Codechange: No need to buffer and format sign strings, they are C-style strings to start with https://git.io/fhAoZ
05:36:29 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN updated pull request #7304: Codechange: No need to buffer and format sign strings, they are C-style strings to start with https://git.io/fhAE9
07:00:12 *** Alberth has joined #openttd
07:00:12 *** ChanServ sets mode: +o Alberth
08:09:55 *** andythenorth has joined #openttd
08:29:14 *** Progman has joined #openttd
08:42:51 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh approved pull request #7305: Codechange: Make std::stack use std::vector container in string formatting/drawing. https://git.io/fhAK5
08:52:23 *** sla_ro|master has joined #openttd
09:06:05 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN merged pull request #7305: Codechange: Make std::stack use std::vector container in string formatting/drawing. https://git.io/fhAVC
09:11:26 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh commented on pull request #7304: Codechange: No need to buffer and format sign strings, they are C-style strings to start with https://git.io/fhA6Z
09:15:31 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on pull request #7304: Codechange: No need to buffer and format sign strings, they are C-style strings to start with https://git.io/fhA6W
09:16:06 <peter1138> Not sure what other GUI lists would be worthwhile messing with.
09:16:43 <peter1138> I don't think any others can be added to/removed, with a similar number of rows, as the sign list.
09:17:33 <nielsm> stations and towns pretty much have to go through FormatString since they have generated names
09:18:26 <peter1138> I'm not sure what that extra FormatString with JUST_RAW_STRING is doing there.
09:18:55 <peter1138> If it's to sanitize input, then it should be sanitized at the input stage, not every time during output.
09:19:01 <peter1138> But I'm not sure it even does that.
09:19:41 <nielsm> I wonder what happens if you put a {NUM} control code byte inside a sign name
09:19:50 <nielsm> (via memory hacking or whatever)
09:20:01 <peter1138> Yeah, quite, if it's memory hackig, who cares.
09:20:38 <peter1138> All those bits in strings.cpp touched in 102f811d027 seem unnecessary.
09:20:40 <nielsm> it's mostly about hacked saved/forged network packets
09:21:13 <peter1138> That should be validated. Should be.
09:21:59 <peter1138> saveload already calls str_validate.
09:22:25 <peter1138> As does network code.
09:23:21 <peter1138> I think it's missing from non-network commands, so maybe that's a thing.
09:24:15 <nielsm> hmm I should try to get hold of the other person(s) translating danish.txt here, I may have Questions about some word choices
09:24:19 <peter1138> Hmm, script API validates in its own DoCommand
09:29:26 <nielsm> edited in a BIG_TEXT code
09:30:25 <peter1138> What about the code for {NUM}?
09:30:46 *** kiwitree has joined #openttd
09:32:22 <peter1138> Is that with it with the current FormatString JUST_RAW_STRING, or with strecpy?
09:33:18 <nielsm> so far NUM is just displaying a 0
09:33:28 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN updated pull request #7304: Codechange: No need to buffer and format sign strings, they are C-style strings to start with https://git.io/fhAE9
09:33:31 <nielsm> not sure if I can force a different value into the dparams
09:33:34 <peter1138> Hmm, ok, so basically it's not there to protect anything.
09:34:56 <nielsm> the NUM is converted to a literal zero in the sign text when I try editing it again
09:35:02 <nielsm> the BIG_TEXT was not converted
09:35:27 <peter1138> that's handled in the display code, not the formatting code.
09:35:46 <nielsm> this may also allow for DoS on other network clients?
09:36:40 <peter1138> How would you insert it?
09:37:06 <peter1138> Input from the network is already sanitized.
09:37:15 <nielsm> hacked client (faking its revision name) and have it prepend a code to a sign name command being transmitted
09:37:42 <nielsm> well if strings received are sanitized it's fine
09:37:55 <peter1138> src/network/core/packet.cpp:311
09:40:32 <nielsm> iow. the only use would be making hacked savegames to play pranks on others (since there shouldn't be risk of RCE either, everything is bounds checked)
09:59:54 *** andythenorth has joined #openttd
10:09:52 <nielsm> Gabda: good idea, I'd personally name the macros something shorter to type, e.g. TTIC/TTOC
10:13:36 <Gabda> yeah, that makes more sense
10:16:52 <andythenorth> it's not needed for gameplay
10:17:05 <andythenorth> it was only added because output cargos were limited to 2
10:18:04 <Gabda> CTIC, CTOC, or TICC, TOCC?
10:21:10 <Gabda> it is my first time seeing the FIRS page, it is really nice
10:21:28 <Gabda> andy, did you do this alone?
10:23:24 <Gabda> I found the repo, and the committers
10:23:49 <andythenorth> I am the primary author, I get essential help from multiple other people :)
10:24:14 <andythenorth> nielsm: categorising is hard :P
10:24:56 <Eddi|zuHause> nielsm: the ways if steam are unknowable...
10:27:04 <Eddi|zuHause> Gabda: do they not work as complete replacement for TIC and TOC?
10:27:31 <nielsm> original TIC/TOC are faster
10:27:44 <nielsm> but give difficult to interpret measurements
10:28:52 <andythenorth> but will it coal->petrochemicals in Fischer-Tropsch plant? o_O
10:28:58 <Gabda> I don't know if others want to use the original one or not, so only adding and not removing seems safer
10:31:30 <Eddi|zuHause> seems odd to leave that decision up to a random person looking for a debug function who won't really know the difference
10:34:21 <Gabda> that could be solved with a more detailed documentation
10:36:36 <Gabda> I just saw the petrol station in FIRS: do you need to fuel the vehicles with coal/petrol?
10:37:03 <Gabda> it would be interesting if you could only start from a coal mine/oil refinery
10:38:18 <Eddi|zuHause> no, that is not a thing an industry NewGRF can do
10:38:55 <LordAro> Gabda: i know it's not your issue as such, but technically speaking variables prefixed with double undeescore are reserved for compiler/standard library
10:39:00 <Gabda> can it be done from a vehicle NewGRF?
10:40:35 <Eddi|zuHause> a vehicle newgrf could alter running costs or stuff, but i don't think it can access any data about cargo delivery
10:41:06 <Gabda> LordAro: well, I can make a PR with fixing these variable names in the original TIC and TOC first
10:41:09 <nielsm> you'd need a GS or something controlling it and offering data to vehicles
10:46:48 <Gabda> if I change the variable names to be ANSI-C compatible, is it a codechange or a fix?
10:47:45 <Eddi|zuHause> it aint a fix if it aint broken :p
10:48:21 <LordAro> Gabda: doesn't need a separate PR, just use the same one
10:49:38 <Gabda> ok, same pr, separate commit then
10:52:04 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro commented on pull request #7086: Change #6173: Update SDL driver to use SDL 2.0 https://git.io/fhAiS
10:55:08 *** andythenorth has joined #openttd
11:02:02 <DorpsGek_II> [OpenTTD/OpenTTD] GabdaZM updated pull request #7306: Add: Chrono based TIC() and TOC() in debug https://git.io/fhA6p
11:03:45 <DorpsGek_II> [OpenTTD/OpenTTD] GabdaZM commented on pull request #7306: Add: Chrono based TIC() and TOC() in debug https://git.io/fhAPT
11:06:49 <Gabda> can you suggest some materials where I can start to understand what are these NewGRF-s exactly?
11:10:53 *** Thedarkb-T60 has joined #openttd
11:13:21 <Gabda> the wiki might be more user friendly :)
11:14:41 * andythenorth wonders what the theme should be for FIRS 'In A Hot Country'
11:14:41 <andythenorth> Steeltown theme is easy :P
11:14:41 <andythenorth> it's iron + steel
11:16:10 <andythenorth> all 3 Basic economies are themed to a region
11:17:42 <andythenorth> the current released version of IAHC is themed as 'everything is delivered to ports, there is not much secondary processing'
11:17:42 <andythenorth> and it's really boring :P
11:19:03 <Eddi|zuHause> well, clearly the ports should work like secondary industries in this case
11:19:49 <Eddi|zuHause> basically making a location restricted secondary industry
11:20:56 <andythenorth> I did wonder about that
11:20:56 <andythenorth> changing the production method
11:22:16 <Eddi|zuHause> the other way would be to make the secondary industries available, but funding only
11:22:27 <Eddi|zuHause> like original saw mill
11:22:30 * andythenorth also considering processing to get FMSP and ENSP
11:22:57 <Eddi|zuHause> so you get an agrarian society at game start, and have to industrialize it
11:23:14 <andythenorth> instead of just ports
11:23:14 <andythenorth> oh, so a 'build an industrial economy' kind of thing
11:23:14 <andythenorth> I always thought that would be more a GS thing, but maybe it can be done in newgrf
11:24:15 <andythenorth> it's an option yes
11:24:15 <andythenorth> I think I'd rather see a GS build things for you, when certain triggers are reached
11:24:15 <andythenorth> but that's not in scope for me :P
11:28:41 <andythenorth> be quite easy to do town cargos
11:28:41 <andythenorth> imported vehicles, electronics, pharmaceuticals
11:28:41 <andythenorth> but town cargos are currently very boring
11:28:41 <andythenorth> esp. with cdist, which is broken
11:31:22 <Eddi|zuHause> could distinguish between western ports (vehicles) and easter ports (electronics)
11:32:37 <Eddi|zuHause> and the poriduction of ports should be really low compared to the fundable secondary industries
11:39:11 *** andythenorth has joined #openttd
11:39:13 <andythenorth> ports that check xy location on map? o_O
11:40:40 <andythenorth> I suspect that port placement is already too fragile to support that :)
11:40:50 <andythenorth> there are problems getting them built
11:41:22 <Eddi|zuHause> make a layout that doesn't have to be placed on shore, but restrict that to map edge?
11:42:29 <Eddi|zuHause> customs seem appropriate
11:42:41 <andythenorth> probably needs a backstop :P
11:43:01 <Eddi|zuHause> what is even a backstop?
11:43:04 <andythenorth> "Your island voted to leave the trading block"
11:43:23 <andythenorth> NewPoliticalDisasters
11:43:29 <andythenorth> what even is a backstop, yes
11:43:40 <Eddi|zuHause> i mean, what was the meaning of the word "backstop" more than 2 years ago?
11:43:45 <Eddi|zuHause> did that word exist?
11:44:24 <Eddi|zuHause> new campain: "backstop or brexstop"
11:44:25 <LordAro> the person standing behind the batter in baseball, i think
11:44:44 <LordAro> clearly should've been wicket keeper
11:47:39 <andythenorth> so why are some coast tiles not coast until I've bulldozed them
11:49:18 <TrueBrain> LordAro: you happen to know why PR#7086 replaces SDL1, and not make SDL1 and SDL2 co-exist?
11:50:04 <LordAro> TrueBrain: because that's what we said to do
11:50:12 <LordAro> SDL2 is plenty mature enough at this point
11:50:47 <TrueBrain> in general, we keep old shit around for years (for not always a good reason); so I was wondering why in this case we remove support for a working driver? (honest question btw)
11:51:16 <LordAro> for the same reason we don't support multiple versions of other libraries?
11:51:30 <TrueBrain> except that in this case SDL1 and SDL2 are llike apple and oranges :D
11:51:53 <TrueBrain> are we sure that all targets support SDL2?
11:52:00 <TrueBrain> (like MorphOS, BeOS, etc)
11:53:22 <TrueBrain> mainly the reason I ask, is that I somehow expected them to co-exist, at least for a while, so we could see what breaks for weird targets we support :)
11:53:36 <TrueBrain> but if this means we can remove MorphOS, I am also fine with that :P
11:54:41 <TrueBrain> so that also for sure kills win9x :)
11:55:00 <TrueBrain> so after that PR we rip out all the code related to it too?
11:55:02 <LordAro> does win9x not use something native?
11:55:13 <TrueBrain> possibly it runs the GDI, indeed
11:55:36 <TrueBrain> but we are okay with that consequence? (BeOS, MorphOS mostly, I guess)
11:55:55 <TrueBrain> (I am just asking the question; I have no preference of opinion on the matter really)
11:56:21 <LordAro> haiku has replaced BeOS, afaict
11:56:58 <LordAro> last release 2001, yeah, i think so
11:57:29 <TrueBrain> honestly, I am still of the opinion: if we don't have a CI for it, we should not support it :P (so I guess I do have an opinion on it :D)
11:57:39 <TrueBrain> anyway, think about it .. dropping SDL1 means we lose some users I guess
11:58:03 <Eddi|zuHause> TrueBrain: clearly we need a fringe-os-ci then :p
11:58:29 <TrueBrain> Eddi|zuHause: fringe-os?
11:59:11 <Eddi|zuHause> TrueBrain: i wasn't going to specify which os, so imagine "fringe" as a placeholder
11:59:28 <TrueBrain> well, we should add a mingw-ci :P
11:59:51 <Eddi|zuHause> we also need a disable-network ci and stuff like that
12:00:08 <TrueBrain> or remove disable-network :)
12:00:16 <andythenorth> 7307 might be a FIRS bug, I don't undertand why it allows WATER_CLASS_NONE in those tiles
12:00:27 <andythenorth> and possibly bulldozing the coast sets WATER_CLASS_NONE true
12:00:30 <TrueBrain> the only OS we "support" disable-network for, is DOS
12:00:59 <Eddi|zuHause> so we need a DOS ci :p
12:01:09 <TrueBrain> that would do both, yes
12:01:12 <TrueBrain> and would also test Allegro
12:01:49 <TrueBrain> (which we only have because of DOS)
12:01:55 * andythenorth wonders if there's a way to inspect tile props in-game
12:02:10 <andythenorth> without debugger to step through the map array
12:04:52 <andythenorth> it's trees I think
12:04:59 <TrueBrain> which you are not showing,ofc :P
12:05:06 <TrueBrain> I see what you did there :D
12:05:17 <andythenorth> so if there are trees on a coast tile
12:05:31 <andythenorth> the land info tool is showing 'trees' not 'coast or riverbank'
12:05:41 <andythenorth> not sure what newgrf is getting for waterclass in that case
12:05:45 <andythenorth> but it's not what it wants
12:05:55 <andythenorth> usually I have trees invisible, because they're crap
12:07:11 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain opened pull request #7308: Codechange: Remove ability for SDL to be dynamically loaded on Windows https://git.io/fhAXL
12:07:20 <TrueBrain> LordAro: I made it into a separate PR, because .. that makes most sense to me :D
12:08:46 <andythenorth> so should a coast tile with trees on it have a waterclass?
12:08:55 <andythenorth> and what is it reporting as newgrf var in that case?
12:13:07 <TrueBrain> (I am asking as I was about to do that, but there might be fuzz in #openttd about it already :P)
12:15:12 <andythenorth> TrueBrain: it's just standard "I try asking again to see if I get a different answer"
12:18:56 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain closed pull request #7292: Change: No need to save AI/GS data when joining a multiplayer server https://git.io/fhNyv
12:18:57 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain commented on pull request #7292: Change: No need to save AI/GS data when joining a multiplayer server https://git.io/fhAXB
12:19:23 <andythenorth> who is the trees-on-coasts newgrf variable expert?
12:19:25 <andythenorth> is it peter1138 ?
12:19:30 <andythenorth> is it Eddi|zuHause ?
12:19:37 <andythenorth> is it TrueBrain ?
12:20:02 <Eddi|zuHause> andythenorth: what seems to be the officer, problem?
12:20:06 <TrueBrain> me? NewGRF expert of ANYTHING? :P Ha! Don't make me laugh :D
12:20:40 <andythenorth> Eddi|zuHause: 7307 above :P
12:22:40 <andythenorth> unless it's an nml bug o_O
12:25:35 <Eddi|zuHause> andythenorth: looks like tiletype MP_TREES cannot have a water class
12:25:56 <TrueBrain> I am looking at some PRs, and honestly wondering why we would want them .. there is also no argumentation what the benefit is :P
12:26:11 <andythenorth> TrueBrain: give links?
12:26:22 <Eddi|zuHause> return IsTileType(t, MP_WATER) || IsTileType(t, MP_STATION) || IsTileType(t, MP_INDUSTRY) || IsTileType(t, MP_OBJECT);
12:26:35 <TrueBrain> join as spectator on a SP?
12:26:45 <TrueBrain> I fail to see the usecase
12:26:50 <andythenorth> there was long long long discussion for it
12:26:57 <TrueBrain> so THAT should be in the PR at the VERY least
12:27:02 <andythenorth> some features get accepted because the argument went on for too lojng
12:27:14 <andythenorth> in fact, we make lots of important decisions that way in the world
12:27:27 <andythenorth> whoever can bore eveyone else into submission wins
12:29:58 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain commented on pull request #7158: Add: Client setting gui.start_spectator https://git.io/fhAX2
12:30:05 <TrueBrain> how is that for a rule? :)
12:31:00 <Eddi|zuHause> andythenorth: on first glance i see no technical reason why MP_TREES should have no waterclass if terrain type is shore
12:31:33 <andythenorth> Eddi|zuHause would you mind commenting on the issue? Or I can paste lines from here if you prefer
12:32:17 <Eddi|zuHause> andythenorth: hold off on that for a moment
12:33:47 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain commented on pull request #7190: Fix #7188: AI instance crash when reloading AI in a server, and an AI… https://git.io/fhAXr
12:35:02 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain commented on pull request #7193: Fix #6468: Load correct version of AI as specified during the time of its save. https://git.io/fhAXo
12:35:28 <michi_cc> TrueBrain: peter's comment.
12:37:01 <TrueBrain> michi_cc: ah; well, I guess he never managed to make it less vague :P Will ask him to either be more specific, or we go with your guts :)
12:37:18 <Eddi|zuHause> "// XXX incomplete" <-- how descriptive
12:37:25 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain commented on pull request #7176: Fix #6633: Cargo monitor industry delivery now accounts for which IndustryID the cargo was delivered to https://git.io/fhAX6
12:37:42 <TrueBrain> Eddi|zuHause: it is obvious, not?
12:38:36 <andythenorth> the funny smell in 7176 is that there's no reference GS implementation to test it against
12:38:43 <andythenorth> and it's unlikely anyone will make one
12:38:59 <andythenorth> so there's no way to check for unexpected consequences
12:39:47 <andythenorth> I really think content API changes should require a reference test with content
12:39:59 <andythenorth> it's a bit onerous, but at least it's then proven
12:42:00 <Eddi|zuHause> andythenorth: so you want to provide a test grf for #7000? :p
12:42:00 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain commented on pull request #7086: Change #6173: Update SDL driver to use SDL 2.0 https://git.io/fhAXy
12:42:31 <andythenorth> Eddi|zuHause: yes I do and I will
12:42:37 <andythenorth> not today probably, family stuff to do
12:43:43 <andythenorth> hmm 'track has catenary' is to cater for multi-mode on 3rd rail track?
12:43:52 <andythenorth> or swiss restaurant cars?
12:44:09 <andythenorth> are we sure frosch wasn't trolling? o_O
12:44:30 <Eddi|zuHause> i'm never sure about that. except with TrueBrain, he's always trolling.
12:46:08 <andythenorth> might as well make the nml patch, and include a test case there
12:46:19 <andythenorth> did anyone start reference.grf yet?
12:46:35 <andythenorth> we have maybe 30 or 40 active nml authors by my guess
12:48:38 <TrueBrain> -<U+FEFF><?xml version="1.0" encoding="utf-8"?>
12:48:38 <TrueBrain> +<?xml version="1.0" encoding="utf-8"?>
12:50:39 <TrueBrain> LordAro: your patch (removing SDL) didn't work :P
12:51:08 <TrueBrain> not possible :D You include a file that you remove :P
12:51:16 <Eddi|zuHause> isn't there a git extension that can ignore BOMs same as \r or \r\n changes?
12:51:47 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7308: Codechange: Remove ability for SDL to be dynamically loaded on Windows https://git.io/fhAXL
12:52:09 <TrueBrain> (I made this a new commit, as I wanted to be sure I did the right thing)
12:52:27 <TrueBrain> especially the #undef part
12:53:59 <LordAro> oh, the projects, of course
12:54:06 <TrueBrain> yeah, but at the bottom
12:54:46 <LordAro> i bet the crashlog one is because windows & case insensitivity
12:54:55 <LordAro> so it's just including SDL.h "twice"
12:55:09 <LordAro> not sure how i missed the other one though
12:55:16 <TrueBrain> but what I did is what we expect, right?
12:56:10 <TrueBrain> hmm .. AP was testing an older version .. that is odd
12:56:19 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7308: Codechange: Remove ability for SDL to be dynamically loaded on Windows https://git.io/fhAXL
13:06:32 <Eddi|zuHause> andythenorth: can you test whether this works at all?
13:31:09 *** stefino has joined #openttd
13:36:52 *** Alberth has joined #openttd
13:36:52 *** ChanServ sets mode: +o Alberth
13:37:10 <stefino> Hi all. I'm trying to make icons of signals but have no idea, how to code it into the game. I have a code like this what defines only signals itself but dont know if it is possible to make its icons. It shows signals real graphics now.
13:41:54 *** frosch123 has joined #openttd
13:53:31 *** Thedarkb-T60 has joined #openttd
14:06:22 *** Beerbelott has joined #openttd
14:08:06 <peter1138> What's wrong with SDL on Windows?
14:08:28 <peter1138> It's kinda pointless, but...
14:08:45 <peter1138> It did made testing my OpenGL stuff easier ;)
14:15:10 <LordAro> peter1138: nothing's wrong with it, it's just dynamically loaded, instead of actually being linked
14:19:12 <TrueBrain> basically: we removed lot of special code; nothing changed
14:20:07 *** supermop_Home_ has joined #openttd
14:21:30 <supermop_Home_> Eddi|zuHause you've had me reading about hydrology off and on over the past few weeks here
14:21:48 <supermop_Home_> mostly stuff I've read before but reading again for fun
14:36:55 <Eddi|zuHause> supermop_Home_: how is that my fault?
14:38:06 <supermop_Home_> talking about procedural river generation etc a while back
14:43:04 <Eddi|zuHause> supermop_Home_: yeah, but i did not come up with that idea
14:44:30 *** andythenorth has joined #openttd
14:45:57 <Eddi|zuHause> supermop_Home_: i'm fairly sure that was andythenorth
14:46:24 *** synchris has joined #openttd
14:47:04 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
14:49:48 <andythenorth> I believe someone has a patch
14:50:08 <andythenorth> peter1138: did we implement lava yet btw?
14:50:24 * andythenorth thinks landscape needs to be more 'interesting'
14:50:31 <Eddi|zuHause> andythenorth: so, did you test #7309 yet?
14:50:39 <supermop_Home_> if trees can be sea, can they prefer to be a certain type of tree
14:51:13 <supermop_Home_> like palms or mangroves, or what ever salt tolerant trees they have in boreal areas?
14:51:52 <Eddi|zuHause> supermop_Home_: i'm awaiting your NewGRF
14:52:15 <peter1138> I don't quite get why they need waterclass but...
14:52:36 <Eddi|zuHause> peter1138: read #7307?
14:52:58 <peter1138> Yes but it doesn't explain much.
14:53:32 <peter1138> Trees can be on "the shore" which I guess means a shope that touches water at sea level.
14:53:41 <peter1138> But it's still a tree tile, not a water tile./
14:53:44 <Eddi|zuHause> peter1138: they visually show a shore sprite
14:56:22 <peter1138> But it's still a tree tile.
14:56:49 <Eddi|zuHause> peter1138: there are a bunch of tiles which are not MP_WATER but have a water class
14:56:51 <peter1138> What is querying the water class of a tree tile? And why?
14:57:41 <Eddi|zuHause> peter1138: i'm not conceptually adding something significantly new
14:58:06 <peter1138> I'm sure you're not. I'm just trying to understand it.
14:58:18 <andythenorth> Eddi|zuHause: 7309 appears to work
14:58:26 <Eddi|zuHause> i don't know what you're missing...
14:58:42 <peter1138> What does it solve. That's what I'm missing.
14:58:55 <peter1138> andythenorth says it works. I dunno what works :p
14:58:59 <andythenorth> I need to rewrite 7307 a bit, strictly
14:59:17 <andythenorth> peter1138 it's for finding coast tiles
14:59:43 <Eddi|zuHause> peter1138: i read #7307 as "my custom NewGRF industry placement check fails, because it tests for WaterClass, but some tiles that show up as shore don't have a WaterClass"
14:59:44 <andythenorth> currently having a tree on it makes it not a coast tile
15:00:03 <peter1138> Eddi|zuHause, sorry, I blame github.
15:00:15 <andythenorth> 7307 would bear some rewriting, but eh
15:00:20 <peter1138> Why can't they make "#7307" in the title clickable and a reference.
15:00:28 <peter1138> Didn't notice it :(
15:00:43 <peter1138> So the real issue is that andythenorth plays with trees hidden :D
15:00:52 <andythenorth> and why do i do that?
15:00:58 <andythenorth> it's because tree generation sucks
15:01:04 <andythenorth> didn't I invent newgrf trees once?
15:01:10 <Eddi|zuHause> we should go through the regression complaints and check whether they are significant
15:01:16 <andythenorth> where's that forests PR, the one that doesn't work?
15:01:47 <peter1138> I probably have a patch for something somewhere.
15:02:00 <peter1138> Maybe I should put ALL my patches up in random PRs. With no context.
15:02:08 <peter1138> I mean, Samu does it :D
15:02:17 <Eddi|zuHause> does that work out for him? :p
15:03:12 <peter1138> I mean, it works far better than leaving them as random .diffs in a directory on a server.
15:03:48 <Eddi|zuHause> peter1138: but please import them into git with commit dates of 2005 :p
15:04:04 <Eddi|zuHause> (or whatever file date :p)
15:04:25 <peter1138> git commit --date does actually exist.
15:06:37 <peter1138> Okay, the regression fails because the game now thinks it's a water tile, so can't build the HQ.
15:06:39 <Eddi|zuHause> so, the regression seems to fail placing a HQ, but the error string is "ERR_TUNNEL_CANNOT_BUILD_ON_WATER"?
15:07:10 <peter1138> IsCoastTile() specifically only returns true if tiletype is MP_WATER.
15:07:27 <Eddi|zuHause> i can look at that
15:08:49 <peter1138> ... can't build on water
15:09:21 <Eddi|zuHause> peter1138: i'm a bit confused about the "TUNNEL" bit
15:10:38 <Eddi|zuHause> ah, regression fixed
15:10:41 <peter1138> src/script/api/script_tunnel.hpp: ERR_TUNNEL_CANNOT_BUILD_ON_WATER, // [STR_ERROR_CAN_T_BUILD_ON_WATER]
15:10:50 <peter1138> That seems to be an API bug.
15:11:08 <peter1138> It maps that string to a tunnel error.
15:13:19 <peter1138> That seems dubious :/
15:14:09 <Eddi|zuHause> ok, i could put in an extra check for water class
15:15:30 <peter1138> Hmm, so plain grass tiles into the sea are actually MP_WATER?
15:15:57 <Eddi|zuHause> well, after they've been flooded
15:21:35 <Eddi|zuHause> now with doc changes
15:22:49 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
15:25:34 <andythenorth> OP just hasn't actually tried it, so assumes I've banned it
15:25:48 * andythenorth would never assume anything without testing first, right? :P
15:26:06 <Eddi|zuHause> no, who in his right mind does that?
15:26:34 <andythenorth> someone showed me ctrl-drag signals last year
15:26:39 <Alberth> ask for a test case :p
15:26:40 <andythenorth> after I whined about needing that feature
15:26:41 <peter1138> Eddi|zuHause, so for the other tile types it doesn't matter that they are not flagged as CoastTiles, becuase... they require explicit clearing. Right?
15:26:46 <andythenorth> exists for about 10 years :P
15:27:01 <Eddi|zuHause> peter1138: yeah, i think so
15:27:23 <Eddi|zuHause> peter1138: at least nobody ever complained about them :p
15:28:21 <andythenorth> ^ found that in Jan this year
15:28:35 <Eddi|zuHause> andythenorth: there's a reason why these Ctrl+whatever features a called "hidden features" :p
15:28:47 <andythenorth> ctrl-click depot flip?
15:28:57 <andythenorth> oh I was getting kbanned if I mentioned that again :(
15:29:37 <Eddi|zuHause> andythenorth: i'd post that picture with the text "wtf are you talking about?!?"
15:30:01 <andythenorth> that helps make a nice community, yes :)
15:30:17 <andythenorth> same OP that wanted all non-newest newgrfs deleted on bananas
15:30:22 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
15:31:10 *** gnu_jj__ has joined #openttd
15:31:24 <andythenorth> Alberth: so tags
15:31:37 <andythenorth> obviously tags can be other formats thatn x.y.z
15:31:47 <andythenorth> e.g. numeric build number or whatever
15:32:06 <Alberth> seems a more clean way out imho
15:32:10 <andythenorth> tends to make a lot of spam in tags, not sure I care though
15:32:20 <Eddi|zuHause> andythenorth: technically, "master" is also only a tag
15:32:23 <andythenorth> it's an interesting approach to try
15:32:52 <Eddi|zuHause> (just one that is automatically moved if you make a commit)
15:33:14 <Alberth> alternatively, you can make a file with hashes to build numbers or so
15:33:20 <andythenorth> so we'd parse existing tags, find the chain of numeric version number tags, and increment?
15:33:41 <andythenorth> what happens if two repos conflict on that?
15:34:06 <Alberth> tags don't get merged afaik
15:34:37 <andythenorth> Eddi|zuHause: context.... "Still pondering on the numeric id. While code can generate something (say "days since 2000" or so), it immediately breaks if you ever release twice on a single day. You can increase precision, at the cost of a (possibly much) longer number."
15:34:37 <Alberth> ie you get "origin/tag" and just "tag"
15:34:38 <Eddi|zuHause> i think you need some special git incantation to pull the tags
15:35:23 <Alberth> yes, tags aren't copied by default
15:35:39 <Alberth> at least light-weight tags
15:35:49 <Alberth> not sure about the not light-weight tags
15:36:13 <Eddi|zuHause> andythenorth: i'm more inclined to do something like "number of commits since root entry"
15:36:19 *** Gustavo6046 has joined #openttd
15:36:37 <Eddi|zuHause> you'll get duplicates on branches
15:36:38 <Alberth> Eddi|zuHause: + branchname to keep it unique
15:36:40 <peter1138> Should the land info tool report these tiles as "trees" or "coast or riverbank with trees" or something?
15:37:12 <Alberth> and horribly breaking if you ever change history
15:37:12 <Eddi|zuHause> Alberth: you can put the branch name into the GRF name, but not in the A14 numeric version
15:37:55 <andythenorth> peter1138: if it can compound, then "coast or riverbank with trees" is better
15:38:10 <peter1138> Well, it'd be an extra string.
15:38:13 <andythenorth> might be TMWFTLB, translations
15:38:23 <Eddi|zuHause> maybe you can add a file on release tags that fixes a version number
15:38:40 <Eddi|zuHause> and make sure the bananas releases are monotonous
15:38:58 <Eddi|zuHause> development releases could stay somewhat fluid
15:39:00 <andythenorth> so what's the version number used for?
15:39:10 <andythenorth> I was a recipient of it rather than an author
15:39:15 <Eddi|zuHause> andythenorth: the game only shows the highest number in the newgrf search
15:39:18 <andythenorth> action 14 uses it to determined compatibility?
15:39:34 <Eddi|zuHause> yes, also compatibility
15:39:37 <Alberth> users use it to find the newest
15:39:54 <Alberth> how many digits can it have?
15:40:18 <andythenorth> strikes me as being quite hg-centric eh :P
15:40:26 <andythenorth> lot of hg users involved in that spec :)
15:40:48 <Eddi|zuHause> i don't think hg was relevant for that spec
15:41:40 <Eddi|zuHause> i read that as "DWORD"
15:42:50 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
15:43:57 * peter1138 tries to decide what to do with the rv-path-cache.
15:44:36 <Eddi|zuHause> NotSaveLoad branch :p
15:47:06 <andythenorth> ok so we do need a numeric rev
15:47:25 <peter1138> manually update it :p
15:47:36 *** snail_UES_ has joined #openttd
15:47:44 <andythenorth> I do manually update the min. version
15:48:01 <Eddi|zuHause> pre-commit-hook?
15:48:30 <andythenorth> can it parse semver?
15:54:29 <andythenorth> switch back to hg? o_O
15:55:22 <peter1138> git rev-list --count too ambiguous?
15:55:41 <andythenorth> it might be fine
15:55:52 <Alberth> add a file with hashes and numbers, or have tags where the compatibility cannot be specified
15:55:58 <andythenorth> per commit, it's coming out with quite different results to hg
15:56:10 <andythenorth> which would be fine if no grfs were released already :P
15:56:17 <Eddi|zuHause> yes, because hg tracks the number across all branches
15:56:18 <andythenorth> I could just add 1 to the start :P
15:56:22 <TrueBrain> what are you trying to solve?! :)
15:56:35 <andythenorth> I have to have a newgrf version number for $reasons
15:56:41 <peter1138> git rev-list HEAD --count I guess.
15:56:43 <Eddi|zuHause> andythenorth: just add the current highest number
15:56:58 <andythenorth> expedient solution
15:57:28 <LordAro> i did that for aroai, iirc
15:57:33 <Eddi|zuHause> (i'm pretty sure i said that last time)
15:58:03 <andythenorth> I was hoping we'd find something less crude
15:58:14 <andythenorth> it's good engineering, but seems wrong
15:58:27 <Alberth> git isn't a very nice vcs :p
15:58:54 <Eddi|zuHause> git is horrible to use, but apparently there is a religious following
15:59:00 <Alberth> it's very technically correct but not nice
15:59:06 <peter1138> It just doesn't try to pretend there's a canonical revision number.
15:59:19 <Alberth> it totally ignores user needs
15:59:30 <andythenorth> I think it's quite divisive
15:59:40 <andythenorth> as soon as I started using git, I could get stuff done
15:59:45 <andythenorth> I never had to learn how it worked
15:59:49 <Eddi|zuHause> i wouldn't choose git if i weren't pressured into it
16:00:28 * andythenorth still hasn't learnt how it works
16:00:54 <Alberth> plenty of knowledgeable people around :)
16:01:03 <andythenorth> I never needed to learn it so far
16:01:30 <andythenorth> ha ha remember svn?
16:01:46 * andythenorth remembers when we moved from whatever predated svn
16:01:46 <Eddi|zuHause> it has gotten better with github, it actually provides a "here's how to do a few common tasks" introduction
16:02:38 <peter1138> andythenorth, you could just tag every release?
16:02:48 <peter1138> Or do you release pretty much every commit?
16:02:58 <andythenorth> I do annotated tags for releases
16:03:17 <Alberth> every push gets build at devzone + nightlies
16:03:25 <andythenorth> but the tag rev there has nothing to do with this tedious version number
16:03:31 <andythenorth> I know why we have to have it
16:03:50 <andythenorth> newgrf has grfid, version, and outside the spec, we have a convention of semver tagging
16:03:59 <andythenorth> it's always looked like a cluster fuck to me, but that's newgrf :P
16:04:24 <Eddi|zuHause> andythenorth: well, you could also live with the fact that nightlies have no meaningful version, and only use release tag version increments
16:04:47 <peter1138> git describe --tag?
16:04:50 <Alberth> could use git describe for that
16:05:12 <andythenorth> I am happy to use whatever crude solution works
16:05:23 <andythenorth> but there are another 100 or devzone projects to port behind FIRS
16:05:43 <andythenorth> eh, maybe those users won't like git either :D
16:05:59 <andythenorth> this seems to be a repeating issue eh?
16:06:07 <andythenorth> didn't we have same problem for OpenTTD?
16:06:35 <peter1138> Although we "fixed" it by not having any merge commits.
16:07:03 <Eddi|zuHause> we didn't fix it, we just ignored the need for an incremental version
16:07:25 <andythenorth> can I do that? :P
16:07:31 <Eddi|zuHause> (newgrf versioning still broken)
16:07:40 <peter1138> $ git describe --tags
16:07:49 <Eddi|zuHause> andythenorth: not really
16:07:52 <peter1138> So... is 3.0.12-4 useful?
16:08:06 <peter1138> Or does it have to be a single number?
16:08:15 <andythenorth> I am unsure what a14 requires
16:08:17 <Alberth> not in the grand scheme of things, where 3.0.13 should have a higher number
16:08:18 <Eddi|zuHause> peter1138: has to be converted to a dword
16:08:51 <andythenorth> LordAro: that XKCD is about hg surely? :P
16:09:00 <andythenorth> that's how I was taught hg
16:09:04 <andythenorth> almost word for word
16:09:36 <andythenorth> a commit hook that pushes to a remote hg repo, over hg-git
16:09:45 <andythenorth> then a script that reads the rev from that hg repo
16:09:50 <peter1138> major << 24 | minor << 16 | build << 8 | revision
16:10:21 <peter1138> 3.0.12-4 becomes 03000C04.
16:10:39 <peter1138> Of course, it's still not unique across branches.
16:10:50 *** drac_boy has joined #openttd
16:10:51 <peter1138> So you might as well just use the rev-list version :p
16:11:02 <drac_boy> hi there .. any interesting codings today? :)
16:11:09 <andythenorth> I wonder if the 'unique across branches' requirement matters, in the real world
16:11:23 <andythenorth> this isn't an algorithmic proof that we need :P
16:11:56 <TrueBrain> force branches to have an unique NewGRF ID? :)
16:12:06 <TrueBrain> (as otherwise it doesn't make sense anyway)
16:12:10 <andythenorth> can I reserve some ID ranges please?
16:12:30 <TrueBrain> branches in NewGRF context make so little sense to me :)
16:12:48 *** Eddi|zuHause2 has joined #openttd
16:12:48 <andythenorth> I have switched to git mostly so I can use branches
16:13:16 <andythenorth> case in point, NRT
16:13:29 <andythenorth> the first public test was a branch of Road Hog
16:13:53 <andythenorth> also, the major problem with testing newgrf spec changes is having grfs
16:13:58 <TrueBrain> and version wise ... it cannot be newer or equal to 'master' ..
16:14:00 * andythenorth plans to use branches for that
16:14:04 <TrueBrain> so .. another NewGRF ID is the only choice there :P
16:14:13 <andythenorth> I don't disagree
16:14:19 <andythenorth> is that like agreeing?
16:14:25 <drac_boy> so anyway just a bit curious about rechecking on this but could one probably made a wagon as a empty sprite (aka its invisible like a cloak) .. it'll just be difficult to deal with in the depot
16:15:26 <andythenorth> but you can use a specific depot sprite if you need to
16:15:30 <Alberth> have a tag as first commit in a branch, and use that as base for naming things?
16:16:09 <Beerbelott> I am testing a new geature which makes me generate new games from heightmap files directly from CLI
16:16:14 <Alberth> no idea if you can find such a tag
16:16:21 <andythenorth> so given the limitations of grf ID, how could I mash the branch name into a grfid?
16:16:25 <Beerbelott> I encounter occasional hang of the game on trying to exit the game
16:16:40 <Beerbelott> Is someone w/ experience available to help me track down that oddity?
16:16:51 <andythenorth> do we have any way to check for grfid colisions?
16:16:51 <Alberth> andythenorth: git describe relative to that tag
16:17:12 <andythenorth> can I apply for a block of grfids?
16:17:16 <andythenorth> like IP v4 addresses
16:17:48 <Alberth> doesn't there exist something like that in the spec?
16:20:20 <TrueBrain> can we remove NewGRF? :)
16:20:48 <andythenorth> yes, if we replace it
16:21:02 <Beerbelott> seems the editor code is buggy...
16:21:02 <andythenorth> we could make a new set of mistakes
16:21:21 <TrueBrain> I recently read a rant about how XML is awesome, we just all use it wrong
16:21:37 <andythenorth> I am tempted to ask for the link
16:21:42 <andythenorth> but my head hurts, and it won't be funny
16:21:50 <drac_boy> ty andy, and yeah I noticed re the ottd-specific depot sprite thing too .. might think about that
16:21:55 <Alberth> as in, don't try to manually read or write it?
16:22:21 <drac_boy> anyway going think some ore re 2-part "wagon" as to be able to carry 2 cargos at once .. still not sure if I'll want to bother releasing the grf like that tho
16:22:29 <andythenorth> imagine it without namespaces :(
16:23:03 <andythenorth> I did wonder about newgrf in yaml
16:23:28 <andythenorth> I think I even wrote a pseudo-code thing for it somewhere
16:23:50 <Alberth> you could get quite far even, I think
16:25:19 <andythenorth> oh yaml can take binaries
16:25:24 * andythenorth didn't know that
16:25:56 <andythenorth> picture: !!binary | [data]
16:26:11 <andythenorth> so it could handle sprites
16:26:19 <Alberth> it also knows about sharing data
16:26:21 <andythenorth> and it can take arbitrary classes
16:26:42 <andythenorth> so we could encapsulate callbacks / switch chains as arbitrary functions, encoded
16:27:04 <Alberth> nice linker format :p
16:27:37 <andythenorth> so make OpenTTD read yaml
16:27:47 <andythenorth> write a utility that helps authors create valid yaml
16:28:30 <Alberth> no, nfo is the assembly language of the processor (ie the stuff inside "binary")
16:28:31 <andythenorth> presumably we could have a compatibility mode to existing grf spec?
16:28:55 <andythenorth> hmm nvm, we talk at right angles :)
16:28:59 <drac_boy> alberth yeah I was just about to say 'no, but you could make it yaml-nfo instead' :)
16:29:03 <andythenorth> I was proposing removing the entire current spec
16:29:16 <andythenorth> lock stock, two smoking barrels
16:29:52 <Alberth> wouldn't that be simply grf version n+1 ?
16:30:04 <Alberth> very non-backawards compatible
16:30:07 <andythenorth> seems versions are hard :P
16:30:54 <andythenorth> maintaining two entire content APIs inside OpenTTD is unwise
16:31:03 <andythenorth> also doesn't GS need removed?
16:33:22 <Alberth> drac_boy: could be nice and hopefully have much more text in it
16:33:56 <andythenorth> I was supposed to be fixing FIRS version number :P
16:33:58 <Alberth> should just ban all numbers :p
16:34:03 <andythenorth> currently I've hard-coded it to '1'
16:34:09 <andythenorth> I mean, it works
16:34:40 <Alberth> that only needs to change when you break compatibility right?
16:37:16 <Alberth> but rev count from a tag near the base of the branch, starting with same name as the grf. Would work if you don't change public commits
16:37:56 <Alberth> don't know what happens if you merge such a thing
16:39:55 *** Wormnest has joined #openttd
16:52:03 <drac_boy> hm..off for today I guess :)
16:52:09 <DorpsGek_II> [OpenTTD/CompileFarm] TrueBrain opened pull request #27: Add: also support CMake projects (this takes precedence) https://git.io/fhAyU
16:52:16 <TrueBrain> anyone wants to press the approve button there ^^ ? :)
16:56:23 <DorpsGek_II> [OpenTTD/CompileFarm] frosch123 approved pull request #27: Add: also support CMake projects (this takes precedence) https://git.io/fhAyk
16:57:03 <frosch123> who is allowed to assign labels to prs?
16:57:16 <frosch123> it looks like people opening them cannot do that
16:57:27 <TrueBrain> I think that is the default, yes
16:57:37 <TrueBrain> only contributors, I am guessing
16:58:01 <frosch123> well, i unsubbed from ottd since i cannot handle 1000 mails a month
16:58:10 <frosch123> but i would still like to find newgrf related prs
16:58:44 <TrueBrain> haha, I unsubbed for the same reason :D
16:59:10 <TrueBrain> not sure if something exist that labels based on the PR changes or keywords
16:59:27 <Alberth> it has an API doesn't it?
16:59:33 <TrueBrain> but currently the PR list is huge, and a bit noisy :)
17:01:37 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
17:02:09 <TrueBrain> owh, right, I wanted to wait with tha tpush till the linux dockers were updated .. oops .. forgot
17:04:34 <DorpsGek_II> [OpenTTD/CompileFarm] TrueBrain merged pull request #27: Add: also support CMake projects (this takes precedence) https://git.io/fhAyU
17:05:02 *** gelignite has joined #openttd
17:07:28 <DorpsGek_II> [OpenTTD/OpenTTD] nikolas updated pull request #7086: Change #6173: Update SDL driver to use SDL 2.0 https://git.io/fhamZ
17:11:30 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro approved pull request #7308: Codechange: Remove ability for SDL to be dynamically loaded on Windows https://git.io/fhAyG
17:11:47 <TrueBrain> LordAro: yes, it counts like that
17:11:50 <TrueBrain> but .. I also looked at it
17:13:08 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain merged pull request #7308: Codechange: Remove ability for SDL to be dynamically loaded on Windows https://git.io/fhAXL
17:13:10 <andythenorth> I was trying to keep up with PRs and maybe label a few
17:13:21 <andythenorth> but the rate is high, and the quality tends to low
17:13:36 <andythenorth> there is only so much I want to be andythenorth doing admin
17:13:43 <andythenorth> I would much rather be andythenorth making newgrfs
17:14:16 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro approved pull request #7304: Codechange: No need to buffer and format sign strings, they are C-style strings to start with https://git.io/fhAyc
17:14:39 <TrueBrain> your own fault; you make 100000 PRs
17:14:50 <TrueBrain> don't be bitching you receive emails because of that ;)
17:16:57 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
17:17:53 <LordAro> i've received 38 OTTD emails today alone
17:18:15 <LordAro> and almost none of them are to do with anything i've done
17:19:17 <LordAro> TrueBrain: if Samu says another thing without any context at all, can you kick him?
17:19:26 *** DorpsGek sets mode: +o LordAro
17:20:45 <TrueBrain> so tempted to 'fix' #7298, just so we can merge it :)
17:21:05 <LordAro> oh, i meant to go shopping
17:21:40 <TrueBrain> "make: *** No rule to make target 'test'. Stop." :D
17:22:24 <Samu> I don't know how to make a templated function
17:22:31 <TrueBrain> w00p, CMake autodetection works on Linux too (on CI) :D
17:23:07 <Alberth> yeah, don't try to configure anything else than default :p
17:23:13 <LordAro> Samu: none of us were born with the knowledge either, dear
17:23:32 <LordAro> TrueBrain: relatedly, i did find myself missing the `run` target
17:26:28 <TrueBrain> LordAro: yeah, and a few more
17:26:35 <TrueBrain> but those can be fixed up later :)
17:26:44 <TrueBrain> those are the easy things to fix :)
17:26:53 <TrueBrain> awh, you cannot define a custom target 'test'
17:27:11 <LordAro> heh, yeah, didn't even think about all the bundles targets
17:28:32 <TrueBrain> but having it all in CMake means we are OS agnostic :D
17:29:12 <Alberth> for the OSes supported by CMake :p
17:30:38 <TrueBrain> fair; but that is the whole idea of the cmake push :)
17:37:33 <TrueBrain> hmm .. 'make test' doesn't compile the binary ..
17:40:49 <Beerbelott> How does one add a commit to an existing PR? does adding a commit to the branch on which the PR is based enough?
17:41:38 <TrueBrain> okay .. 'make test' in CMake does not run 'make all' first
17:41:58 *** Wormnest has joined #openttd
17:44:16 <DorpsGek_II> [OpenTTD/CompileFarm] TrueBrain opened pull request #28: Fix: CMake 'make test' does not run 'make all'. Make this explicit. https://git.io/fhAyP
17:45:09 <TrueBrain> sometimes 3 letters are enough to answer a question :)
17:45:19 <TrueBrain> other times it seems you need to be a wizard :P
17:49:40 <Samu> how do I download older versions of my script from bananas?
17:49:46 <Samu> I thought I could do that :(
17:50:13 <Samu> isn't there an archive website?
17:50:49 <TrueBrain> by ToS of BaNaNaS, you can only download the latest version
17:50:56 <TrueBrain> (people never read a ToS :( )
17:52:08 <nielsm> did anything of note happen in the last 7 hours?
17:53:38 <Alberth> probably, but possibly not in this channel
17:55:02 <Samu> plz i need several versions of my ai to test my PR
17:55:03 <TrueBrain> nielsm: CMake is starting to work more and more :P
18:01:24 <TrueBrain> nielsm: I think LordAro was bitching you misspelled the word :)
18:03:04 <DorpsGek_II> [OpenTTD/CompileFarm] TrueBrain merged pull request #28: Fix: CMake 'make test' does not run 'make all'. Make this explicit. https://git.io/fhAyP
18:03:55 <Samu> rip LuDiAI AfterFix v1-v8
18:04:03 <Samu> i thought i could download older vers
18:04:43 <TrueBrain> glx: I think I can run regression via MSVC too, via CMake \o/ :D
18:04:47 <TrueBrain> this will be so goooooodddddd
18:05:01 <TrueBrain> just have to figure out how to run a project inside a solution :D
18:07:22 <glx> I think I tried to do that, and ended with running regression directly in CI yaml
18:08:20 <TrueBrain> and I managed to fix lzo shit
18:08:28 <TrueBrain> many OSes fix up lzo to be less shit
18:08:47 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN merged pull request #7304: Codechange: Sort and filter sign names directly rather than through string system. https://git.io/fhAE9
18:08:54 <TrueBrain> their include files include <lzo/lzoconf.h> .. and yes, <> and lzo/ ...
18:09:03 <glx> yes lzo provided .pc is broken
18:09:03 <michi_cc> peter1138: Did you have any further thoughts on #7176?
18:09:14 <TrueBrain> glx: well, yes and no .. for most OSes, it works fine :)
18:09:20 <TrueBrain> and it is more like all other libs do stuff
18:09:28 <TrueBrain> most OSes fix the headers
18:09:30 <peter1138> michi_cc, not really. I wasn't sure of the intentions of it all when I first looked.
18:09:35 <TrueBrain> it is really the headers that are wrong
18:09:38 <peter1138> michi_cc, I'm still not sure but if it works... :)
18:09:45 <TrueBrain> the rest is how it "should be"
18:10:08 <TrueBrain> but brew for example does not fix them
18:10:13 <TrueBrain> causing this ... misery
18:10:44 <peter1138> Samu, ^^ that last merge.
18:11:21 <glx> brew does weird stuff, a separate dir for installed lib is not optimal
18:11:30 <TrueBrain> yeah, but that is not really the issue here
18:11:34 <TrueBrain> it is just that lzo is foobar
18:11:38 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh commented on pull request #7306: Add: Chrono based TIC() and TOC() in debug https://git.io/fhASm
18:11:58 <michi_cc> I'm sure there is at least one CityBuilder server out there that relies on the wrong behaviour, but bug-compatibility isn't a good goal...
18:12:16 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on pull request #7304: Codechange: Sort and filter sign names directly rather than through string system. https://git.io/fhASY
18:13:01 <peter1138> michi_cc, I got confused wondering about when it is source and when it is destination.
18:13:04 <glx> #include <lzo/lzo1x.h> was failing in our code too IIRC
18:13:18 <TrueBrain> glx: yup; most other libraries all don't require the 'lzo' part (like freetype, SDL,..)
18:13:32 <TrueBrain> so like 4 mistakes in a row results in a very small solution space to fix this
18:13:38 <TrueBrain> but okay .. it is fixed, it works :)
18:14:01 <peter1138> Samu, game should still be playable with the sign list open now.
18:14:37 <Samu> btw I need old versions of my AI, or of someone's AI
18:14:39 <DorpsGek_II> [OpenTTD/OpenTTD] michicc approved pull request #7176: Fix #6633: Cargo monitor industry delivery now accounts for which IndustryID the cargo was delivered to https://git.io/fhASs
18:14:48 <DorpsGek_II> [OpenTTD/OpenTTD] michicc merged pull request #7176: Fix #6633: Cargo monitor industry delivery now accounts for which IndustryID the cargo was delivered to https://git.io/fhS1x
18:15:37 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
18:15:39 <peter1138> Samu, can't you download your own AIs from bananas?
18:15:48 <Samu> i can only download latest version
18:15:59 <nielsm> okay let's get to derail some valleys
18:16:03 <peter1138> Even from the management-side?
18:16:12 <TrueBrain> nielsm: that is just a cool game :)
18:16:12 <peter1138> Hmm, should I buy it?
18:16:17 <peter1138> I played the demo...
18:16:49 <peter1138> I wonder if they fixed time. In the demo time went super fast, which is a bit awkward for a VR game with basically real time gameplay.
18:18:30 <peter1138> See, you should really have all your versions available as git tags in your AIAI repo ;)
18:19:37 <glx> hmm forgot to add a file TrueBrain ?
18:19:50 <TrueBrain> I love these blank statements :)
18:19:54 <TrueBrain> a bit more specific would be nice :)
18:20:27 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
18:21:05 <Samu> ok, gonna test sign list stuff instead
18:21:22 <TrueBrain> "Unable to retrieve file content for path /azure-pipelines/templates/linux-build.yml from repository self using ref refs/pull/7270/merge and commit 8fc61b5df7913d8e1d1d9f7d6774846a20642ae9: Bad credentials"
18:21:25 *** Eddi|zuHause2 is now known as Eddi|zuHause
18:21:45 <peter1138> Samu, well it's merged now, it better be better.
18:21:55 <peter1138> Samu, it certainly was for me.
18:23:43 <andythenorth> did we delete newgrf yet?
18:24:05 <TrueBrain> no, but the Linux CI is now building via CMake; w00p :)
18:24:50 <Eddi|zuHause> <TrueBrain> so .. another NewGRF ID is the only choice there :P <-- maybe announce that you're reserving 1 million GRF IDs for every possible branch you might do in the future? :p
18:25:20 <TrueBrain> Eddi|zuHause: sounds good to me!
18:26:52 <peter1138> Especially OzTrans'?
18:26:58 <Eddi|zuHause> andythenorth: different GRF-IDs for master (testing) and releases doesn't sound like a hugely terrible idea. that way you can make nightlies available on bananas and hide them, while keeping the latest release there as well
18:27:07 *** andythenorth has joined #openttd
18:28:49 <DorpsGek_II> [OpenTTD/OpenTTD] andythenorth commented on pull request #7147: Change: Synchronize randomness in vehicle introduction… https://git.io/fhASw
18:30:10 <andythenorth> TrueBrain: shall we apply the 'unlucky' hammer to some PRs?
18:30:29 <Eddi|zuHause> don't we have stalebot for that? :p
18:31:05 <andythenorth> we seem to be gaining a lot of optional settings
18:31:17 <TrueBrain> I noticed people push rebases, which voids stalebots activities
18:31:47 <andythenorth> like, how many settings do we need?
18:31:51 <peter1138> Settings are not bad when they make fundamental gameplay changes.
18:32:25 <Samu> hmm generating the list for the first time is taking 4 secs
18:32:39 <Samu> used to be faster if I recall
18:34:07 <Samu> are you sure you fixed it? i'm getting 7,500 ms stall
18:34:22 <peter1138> Of course I tested it.
18:34:25 <peter1138> I posted results too.
18:34:35 <peter1138> There's no way it can be slower :/
18:34:54 <peter1138> It's doing less under all circumstances.
18:36:32 <peter1138> That's from Friday.
18:36:40 <peter1138> You haven't updated/compiled.
18:38:33 <Samu> i fail at this stuff yet
18:38:52 <peter1138> You don't need to push. You pull (or ff merge into your master)
18:39:04 <peter1138> andythenorth, was there a test case for 7147?
18:41:18 <Eddi|zuHause> maybe the default trucks?
18:41:20 *** Thedarkb-T60 has joined #openttd
18:42:55 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on pull request #7147: Change: Synchronize randomness in vehicle introduction… https://git.io/fhAS7
18:43:57 <peter1138> I didn't test beyond 38k signs.
18:44:13 <Samu> wonder what's like when adding signs
18:44:27 <peter1138> Samu, exactly the same.
18:44:42 <peter1138> It still GSort's the entire list.
18:45:07 <peter1138> Samu, also you can have the sign list shaded and it doesn't re-sort now.
18:45:18 <peter1138> Hmm, I'm up to 25ms with 62k signs.
18:45:23 <peter1138> I guess my PC is just a bit faster.
18:47:08 <peter1138> I'm sure it's not useful to have that many signs, but fixing it being horribly slow is worth while.
18:49:14 <Eddi|zuHause> Gwyd voiced a concern about not synchronizing vehicles between different GRFs, i should probably adress that
18:49:30 <andythenorth> peter1138: 7147 can be tested with Iron Horse 2 Alpha 7
18:49:34 <andythenorth> or I'm testing it now
18:49:34 <peter1138> Hmm, what if it's an add-on set?
18:49:48 <andythenorth> it's synchronised anyway, no?
18:49:58 <peter1138> andythenorth, I'm guessing maybe it shouldn't be.
18:50:03 <andythenorth> yeah no it should
18:50:10 <andythenorth> global adjustment of specific days
18:50:12 <Eddi|zuHause> more likely it shouldn't than it should.
18:50:20 <peter1138> For independent sets, does it matter?
18:50:39 <andythenorth> but $somebody is going to care about it being unsynchronised
18:50:45 <Eddi|zuHause> i'd just XOR the seed with the grfid
18:50:47 <andythenorth> $nobody is going to care about it being synchronised
18:51:04 <andythenorth> most people will never even know this exists
18:51:05 <peter1138> andythenorth, well, Gwyd is $nobody?
18:51:15 <Eddi|zuHause> just looking for a way to access that
18:52:22 <andythenorth> $somebody will request synchronisation
18:52:36 <andythenorth> $nobody will request isolation of randomisation per grf
18:53:06 <Eddi|zuHause> peter1138: yeah, i would have found that in the near future :p
18:53:08 <peter1138> Ah, you mean nobody will notice it if it's synchronized, but they will if it's not.
18:54:00 <peter1138> So... bitstuffing into a std::deque, or use two std::deque, or rewrite the saveload stuff to work with a struct in a single std::deque?
18:54:57 <peter1138> Hmm, damn, I pulled added a reference to JGR's patchpack, and now gitg's default "All commits" view starts at svn r5862...
18:55:50 <Eddi|zuHause> i wouldn't do a hacky addon-rewrite, it's not worth it. and a complete rewrite is probably too extensive. so i'd lean towards the two queues
18:56:23 <peter1138> It was a hacky rewrite, although it has precedent in other objects.
18:57:09 <peter1138> JGR's pack allows 16k x 16k maps, so that'd break.
18:57:15 <DorpsGek_II> [OpenTTD/OpenTTD] andythenorth commented on pull request #7147: Change: Synchronize randomness in vehicle introduction… https://git.io/fhA9k
18:57:22 <peter1138> "(The NewGRF debug inspection window is disabled for all map coordinates longer than 27 bits)."
18:57:32 <DorpsGek_II> [OpenTTD/OpenTTD] nikolas updated pull request #7086: Change #6173: Update SDL driver to use SDL 2.0 https://git.io/fhamZ
18:58:35 <Eddi|zuHause> yeah, the bitstuffing is probably not the right way...
18:58:51 <peter1138> Well, I could bitstuff into a uint64 :-)
18:59:20 <peter1138> Maybe I should look into the supporting a struct inside the dequeu.
18:59:57 <peter1138> SLE_UINT32 -> SLE_SUB_DESC
19:00:05 <peter1138> add another field with a _sl_desc
19:00:20 <peter1138> All for this one feature, and it's cache, lol.
19:00:22 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z updated pull request #7147: Change: Synchronize randomness in vehicle introduction… https://git.io/fhDkW
19:01:21 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z commented on pull request #7147: Change: Synchronize randomness in vehicle introduction… https://git.io/fhA9Y
19:01:24 <peter1138> Incidentally, why the heck do we bit-stuff the saveload description stuff :/
19:01:45 <Eddi|zuHause> hysterical raisins?
19:02:53 *** andythenorth is now known as Guest2214
19:02:54 *** andythenorth has joined #openttd
19:03:25 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z commented on pull request #7147: Change: Synchronize randomness in vehicle introduction… https://git.io/fhA93
19:04:33 <peter1138> void *address... I could abuse that :p
19:04:58 <Eddi|zuHause> no, that's the offset to the struct members...
19:05:30 <Eddi|zuHause> i'd rather not touch that :)
19:06:46 <Eddi|zuHause> maybe we should also not synchronize between vehicle types
19:08:43 <Samu> Not a fan of this "magic" -2 value. What's wrong with using versionParam "as is" ? It's documented as "The version of the game script, or -1 if you want the latest"
19:08:57 <Samu> LordAro, where is the document sayin that?
19:09:07 <Samu> i rather fix the documentation
19:12:21 <LordAro> while i may have written that line, i've no clue which one of your many PRs it was on, and i can't be bothered to go and look for it
19:12:41 <andythenorth> so Mac OS includes a password generator in the keychain
19:12:51 <andythenorth> it *regularly* does stuff like this :P
19:13:22 <andythenorth> so either the English language is *crammed* with rude sounding word pairs
19:13:27 <andythenorth> or it's coded by a troll
19:13:40 <Samu> the problem is... in console you can startai "ainame".version
19:13:58 <Samu> when saving, the name of the AI is saved like that
19:14:50 <LordAro> Samu: and please respond to github comments on github
19:15:10 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z updated pull request #7147: Change: Synchronize randomness in vehicle introduction… https://git.io/fhDkW
19:15:36 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z commented on pull request #7147: Change: Synchronize randomness in vehicle introduction… https://git.io/fhA9B
19:16:12 <andythenorth> Azure needs an MS account
19:16:17 <andythenorth> and MS account needs email verification
19:16:36 <DorpsGek_II> [OpenTTD/OpenTTD] SamuXarick commented on pull request #7193: Fix #6468: Load correct version of AI as specified during the time of its save. https://git.io/fhA90
19:16:52 <peter1138> Oh right, that won't work due to templates.
19:17:26 <Beerbelott> Eddi|zuHause: Does the commit checker on PR #2155 fails because of commit names again?
19:18:43 <Eddi|zuHause> Beerbelott: that doesn't sound like a valid PR#
19:19:39 <Beerbelott> PR #7286 (issue #2155)
19:20:09 <andythenorth> ok so I'm in Azure
19:20:10 <Eddi|zuHause> Beerbelott: regression failed
19:20:58 <Eddi|zuHause> Beerbelott: type "make regression" to run it locally
19:21:26 <TrueBrain> andythenorth: good luck!
19:22:54 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
19:22:56 <Eddi|zuHause> <peter1138> Oh right, that won't work due to templates. <-- yeah, the whole whatever_desc need to be replaced by templates, that's why a halfassed rewrite won't work
19:23:04 <Samu> can't find anything about -1, -2
19:23:08 <Beerbelott> Eddi|zuHause: Thx. Oh damn, what have I done? ;) Checking
19:23:27 <TrueBrain> ramblings of a mad man
19:24:45 <TrueBrain> seriously, AP is being annoying today
19:24:53 <LordAro> Samu: i was referring to the function documentation - the /** ... */ comments. in this case, game_scanner.cpp & ai_scanner.cpp
19:25:51 <andythenorth> it is what it is I guess
19:25:56 *** Thedarkb1-T60 has joined #openttd
19:25:59 <andythenorth> needs to d/l all deps all jobs
19:28:39 <andythenorth> oh now I have 2 pipelines
19:28:50 <andythenorth> is that intended?
19:28:56 <TrueBrain> did you want to make 2?
19:29:20 <andythenorth> I went back and edited my first pipeline
19:29:22 <TrueBrain> there is your answer :)
19:29:25 <Samu> darn, i really needed my old AIs for testing
19:29:36 <TrueBrain> saving should not create a new one
19:29:37 <andythenorth> maybe I can edit this yaml in git
19:29:45 <andythenorth> I am shit at the Azure UI
19:30:23 <TrueBrain> like said before, I found it easiest to use the UI at first, as for the YAML you need to know each tasks and what it can do .. I still first make the task via the UI, than export to YAML to see what it should be :P
19:30:36 <TrueBrain> (for that I made a non-YAML-based pipeline)
19:30:44 <andythenorth> I can't find any click-click UI
19:30:45 <TrueBrain> at the bottom when creating a pipeline you can select: Empty Pipeline, I think
19:30:52 <andythenorth> I got a wizard that told me to make a python yaml
19:32:14 <TrueBrain> okay, that is hidden :D
19:32:23 <andythenorth> ok I deleted my project and start again
19:32:26 <andythenorth> that one was rekt
19:32:34 <Eddi|zuHause> andythenorth: since you're now an azure pipeline expert, can you fix extracting the actual error message from the CI build process? :p
19:32:41 <TrueBrain> Builds -> New -> New Build Pipeline -> Use visual editor (!) -> GitHub -> Continue -> Empty Pipeline (at bottom)
19:32:52 <TrueBrain> after that, you have an empty pipeline which can you edit via the UI
19:32:59 <TrueBrain> but that Use visual editor step is .. euh . .yeah .. hidden
19:33:00 <andythenorth> I'll try from scratch
19:33:04 <andythenorth> I was stuck in something with no options
19:36:08 <peter1138> andythenorth, #7147 still gives expected result?
19:37:26 <DorpsGek_II> [OpenTTD/OpenTTD] SamuXarick commented on pull request #7193: Fix #6468: Load correct version of AI as specified during the time of its save. https://git.io/fhA9b
19:37:35 <Samu> Sorry, I'm stuck, can't do anything
19:38:13 <andythenorth> how do I trigger a job on AP?
19:38:15 <Samu> i'm being denied access to my stuff
19:38:17 * andythenorth should read the docs
19:38:47 <peter1138> If you have a savegame that uses an older version, you can download it from that.
19:39:16 <TrueBrain> andythenorth: the bug Queue button :D
19:39:20 <TrueBrain> (and Save & Queue button)
19:40:39 <andythenorth> peter1138: yes 7147 works still
19:41:12 <peter1138> I wonder if the GRFID part is good or bad.
19:41:40 <peter1138> Make the code download the old version :p
19:41:58 <peter1138> You've basically made a patch to load the correct version of an AI.
19:42:06 <peter1138> But you don't actually have the correct version of the AI.
19:42:42 <LordAro> today Samu learns the value of version control systems
19:43:37 <andythenorth> so I can write an AP yaml to delete my own repo? o_O
19:43:58 <TrueBrain> andythenorth: sure, if you like
19:44:33 <TrueBrain> hmm .. what is this about OSX and iconv
19:45:00 <LordAro> TrueBrain: i fixed that a few months ago :p
19:45:11 <TrueBrain> LordAro: fixed what?
19:45:26 <TrueBrain> I love it when people tell me something is fixed while I didnt even know it was broken :D
19:45:44 * andythenorth reading the docs for AP
19:45:51 <andythenorth> it needs to trigger on git push
19:45:51 <LordAro> not all that helpful for your cmake stuff, perhaps
19:45:51 <Beerbelott> Eddi|zuHause: Is it normal 'make regression' faild w/ 'Failed to find a graphics set. Please acquire a graphics set for OpenTTD.' when the OpenGFX set is in ~/.openttd/content_download/baseset/ & not ~/.openttd/baseset/?
19:45:53 <TrueBrain> not sure how that helps me LordAro :)
19:46:02 <TrueBrain> I was more wondering why the fuck iconv is needed for OSX and only for OSX :)
19:46:21 <peter1138> iconv isn't needed on OSX?
19:46:23 <Eddi|zuHause> Beerbelott: that doesn't sound right
19:46:35 <TrueBrain> andythenorth: under Triggers you can enable these things
19:46:40 <Beerbelott> Hmm lemme try on trunk then
19:46:44 <TrueBrain> log 1 "checking iconv... not OSX, skipping"
19:46:49 <peter1138> Beerbelott, it is normal.
19:46:50 <TrueBrain> peter1138: config.lib tells a different story :P
19:47:02 <TrueBrain> no clue what it does :D
19:47:06 <peter1138> TrueBrain, oh. I guess I misread it.
19:47:19 <peter1138> Beerbelott, I don't think it's meant to happen, but it happened for me too.
19:47:23 <Eddi|zuHause> Beerbelott: is it maybe picking up a wrong openttd.cfg?
19:47:36 <DorpsGek_II> [OpenTTD/OpenTTD] nikolas updated pull request #7086: Change #6173: Update SDL driver to use SDL 2.0 https://git.io/fhamZ
19:47:38 <TrueBrain> it is a HUGE function to check for iconv, and only OSX .. what happens if I don't do that ... :P
19:47:39 <peter1138> Eddi|zuHause, it's picking a custom config specifically for the regression test.
19:48:25 <TrueBrain> and why does linux not need it .. hmm
19:48:44 <andythenorth> so its built a grf somewhere
19:48:59 <andythenorth> so now I need a deploy step?
19:49:15 <TrueBrain> you need to publish the result somewhere indeed
19:49:19 <TrueBrain> this can be an artifact at first
19:49:24 <TrueBrain> but that means it is only available on AP
19:49:35 <peter1138> TrueBrain, it's for filename conversion.
19:49:36 <TrueBrain> you can extend it with a Release step, which uses that artifact to do what-ever
19:49:54 <DorpsGek_II> [OpenTTD/OpenTTD] nikolas updated pull request #7086: Change #6173: Update SDL driver to use SDL 2.0 https://git.io/fhamZ
19:50:03 <peter1138> I wouldn't think OSX and non-OSX behave differently there.
19:50:39 <Samu> do you have all your ais from v1 to latest that could be downloaded?
19:50:43 <TrueBrain> peter1138: yeah, I was assuming the same ..
19:51:05 <peter1138> Samu, I've got v7 of AfterFix here
19:51:11 *** Supercheese has joined #openttd
19:51:12 <Beerbelott> Strange, since starting the game from /bin/openttd on CLI never failed like that
19:51:29 <Beerbelott> I wonder how the plumbing for 'make regression' is made...
19:51:48 <Beerbelott> sry bin/openttd, relative to git repo
19:52:49 <TrueBrain> I am back in 2007; even there we only detect iconv for OSX .. and skip it for all other targets ..
19:53:23 <Samu> function MinVersionToLoad() {
19:53:33 <Samu> version 9 has a min of v8
19:54:57 <Beerbelott> for regression, check, the game is called w/ ./openttd -x -c ai/regression/regression.cfg
19:55:56 *** andythenorth has joined #openttd
19:56:48 <Beerbelott> Eddi|zuHause: Well I do not see why the cfg should be a problem, as directories discovery shall be done against env variables, esp. for home folder were .openttd is to be searched for
19:57:20 <glx> because '##[error]build\ZERO_CHECK.vcxproj(0,0): Error MSB4057: The target "RUN_TESTS" does not exist in the project.'
19:57:34 * andythenorth needs to get an Azure Artifacts License somehow
19:57:37 <glx> it exists but is not selected for building, so not a valid target
19:57:52 <andythenorth> "You need a Azure Artifacts license to go further"
19:57:57 <Eddi|zuHause> Beerbelott: the search order is described in section 4.2 in the readme
19:58:01 <andythenorth> "an" MS, "an", not "a"
19:58:07 <Beerbelott> I wonder if the problem does not comme from the fact the binary is called from ./bin/ w/ ./openttd rather than from ./ w/ bin/openttd
19:58:41 <Beerbelott> Eddi|zuHause: I know about that. Game runs fine outside of the regression tests script
19:59:09 <TrueBrain> glx: I am not sure that is the error; but I would rather look how we can enable it, than work around it :)
19:59:18 <TrueBrain> workarounds often result into more issues, in my experience :)
19:59:26 <andythenorth> right who wants to learn AP with me?
19:59:48 <Eddi|zuHause> Beerbelott: does it work if you move ~/.openttd to ~/.local/share/openttd?
19:59:57 <andythenorth> it's not really acceptable that TB is the only one who understands it
20:00:03 <TrueBrain> andythenorth: you can use the PublishArtifact task without license
20:00:14 <TrueBrain> andythenorth: that is also not the case, as I based my work on what nielsm had done :P
20:01:40 <milek7> i set up azure pipelines recently, it didn't want any license
20:01:57 <andythenorth> I'm just going by what the Azure UI tells me
20:02:25 <andythenorth> I'll work around it
20:02:30 <andythenorth> it's probably just broken
20:02:35 <milek7> i didn't use UI, azure ui are generally crap
20:02:57 <andythenorth> the docs are also very confusing
20:03:07 <andythenorth> I think I'll just cargo-cult from other things
20:03:43 <TrueBrain> andythenorth: yes, Artifacts are a bit different
20:03:53 <TrueBrain> the task PublishBuildArtifact is .. not an Artifact on that level
20:03:55 <TrueBrain> but is one on a build level
20:03:58 <TrueBrain> which is confusing as fuck
20:04:00 <TrueBrain> but is what happens :P
20:04:01 <andythenorth> instead of waiting for the job to build all the deps every time
20:04:14 <andythenorth> can I have a job that sits in a loop, polling git?
20:04:22 <andythenorth> and building when it finds a push
20:04:32 <TrueBrain> they are still working on a "cache", to store these things
20:04:33 <andythenorth> do they time-limit jobs?
20:04:37 <TrueBrain> but .. it is not done
20:04:42 <andythenorth> probably quite hard
20:04:46 <andythenorth> invalidating that cache
20:04:48 <TrueBrain> and it makes you a bad neighbour
20:05:05 <andythenorth> during normal use the latency is fine
20:05:10 <andythenorth> but for writing the yaml :(
20:05:20 <andythenorth> it's worse than waiting for nml even
20:05:35 <TrueBrain> welcome to my life, I guess :P
20:07:04 <peter1138> nielsm, that's all your loops sorted, and some wrong paths too.
20:07:06 <andythenorth> TrueBrain: I remember watching that :P
20:07:43 <andythenorth> I could add a virtualenv to the repo
20:07:46 <andythenorth> with all the deps
20:07:52 <andythenorth> it's *really* bad practice
20:08:34 <Beerbelott> Eddi|zuHause: Nope
20:08:34 <peter1138> Build has failed, of course.
20:08:38 <DorpsGek_II> [OpenTTD/OpenTTD] nikolas commented on pull request #7086: Change #6173: Update SDL driver to use SDL 2.0 https://git.io/fhAHl
20:09:12 <TrueBrain> andythenorth: I am a bit lost what you are trying to solve
20:09:23 <Eddi|zuHause> Beerbelott: then i'd try symlinking the baseset dir into the directory where the regression .cfg lies
20:09:28 <TrueBrain> yeah ... I would suggest you just deal with it
20:09:30 <milek7> yay, i bought windows codesigning cert and there are less scary warning signs
20:09:33 <milek7> it feels like extortion business
20:09:38 <TrueBrain> optimizing that rarely results in proper maintainable things andythenorth :)
20:10:38 <andythenorth> AP's 404 is frigging huge font
20:10:49 <andythenorth> and none of the paths I guessed for the artifacts were correct :)
20:11:45 <TrueBrain> andythenorth: you can get the Artifact by clicking on the job
20:11:48 <TrueBrain> a button Artifact appears
20:11:52 <TrueBrain> which allows you to download it
20:12:25 <andythenorth> I found it in 'Summary'
20:12:58 <andythenorth> let's see if it worked
20:13:23 <andythenorth> well it worked though
20:13:42 <andythenorth> ok so now just need somewhere to publish to
20:13:50 <TrueBrain> yeah ... and that part is the most tricky, I guess
20:13:59 <TrueBrain> GitHub releases is an easy first step
20:14:03 <andythenorth> so my lifelong problem with GI deploys
20:14:07 <andythenorth> how to secure the creds?
20:14:11 <TrueBrain> figuring out musa might be a bit much ;)
20:14:12 *** Smedles has joined #openttd
20:14:28 <TrueBrain> in AP you can set variables as secret
20:14:42 <TrueBrain> but in this case, you want to make a Service Connection on project level
20:14:45 <TrueBrain> to for example GitHub
20:14:45 <andythenorth> so they're stored in the pipeline, not my project repo?
20:14:54 <TrueBrain> after that, you can use the Service Connection by name
20:15:02 <TrueBrain> so nowhere in the project is any hint to which secret is even used
20:15:30 <TrueBrain> (and tasks request the token from the Service Connection; not possibly to intercept unless you use an untrusted Task)
20:15:50 <TrueBrain> it is almost like they thought about it :D
20:16:10 <TrueBrain> if you want to publish on GH, you have 2 choices: in the CI chain, or as a Release project
20:16:16 <TrueBrain> I have done Releases, as that is the CD part
20:16:19 <TrueBrain> and I like clear separation
20:16:24 <TrueBrain> but that means it is not in a YAML file in the repository
20:16:47 <andythenorth> hmm I have no clear picture on this part
20:18:12 <andythenorth> there's some way to publish to github?
20:18:13 <Wormnest> Samu: yes I check in here sometimes :)
20:18:25 <TrueBrain> andythenorth: search for GitHub in tasks :)
20:19:17 <andythenorth> is that in Project Settings?
20:19:54 <TrueBrain> so you managed to find all these other tasks
20:19:59 <TrueBrain> yet you don't know what tasks are :P
20:20:14 <andythenorth> there's nothing called tasks in current UI
20:20:19 <TrueBrain> yes, it is full of tasks :)
20:20:20 <andythenorth> tbh the UI is confusing AF
20:20:33 <TrueBrain> just because it is new to you, doesn't mean it is confusing :D
20:20:45 <TrueBrain> a task is like PublishBuildArtifact
20:20:59 <TrueBrain> if you add those via the UI, if you did what I suggested with the Visual Editor
20:21:02 <TrueBrain> you can search for tasks
20:21:09 <TrueBrain> if you did something else ... I cannot really help you :)
20:21:13 <andythenorth> I couldn't find the visual editor :)
20:21:18 <TrueBrain> I told you exactly where it was
20:21:35 <TrueBrain> I could keep myself from making an animated gif :P
20:21:42 <TrueBrain> but really, you want to use the visual editor at first
20:21:49 <TrueBrain> I still use it to make new pipelines
20:21:57 <TrueBrain> doing it in yaml is like going in neck-deep ;)
20:22:05 <andythenorth> ok I delete my project and try again
20:22:18 <TrueBrain> you can make many many projects btw, even on the same GitHub repo ;)
20:22:36 <Beerbelott> Eddi|zuHause: Well the obvious 'solution' was OpenGFX-0.5.2.tar -> ../content_download/baseset/OpenGFX-0.5.2.tar in .openttd/baseset
20:23:30 <Beerbelott> that seems to work... but regression still fails :(
20:23:37 <Beerbelott> ERROR: The script died unexpectedly.
20:23:37 <Beerbelott> Makefile:173: recipe for target 'regression' failed
20:24:31 <andythenorth> ok I found the visual editor now
20:24:37 <andythenorth> I'm not really very good with visual editors
20:25:03 <andythenorth> does anyone know how github pages works?
20:25:15 <andythenorth> I tried using it previously, but it wanted docs committed in the repo
20:25:25 <andythenorth> wondering if there's another way
20:26:00 <TrueBrain> but it is Jekyll :P
20:26:07 <andythenorth> hmm, there's a plugin that commits docs to a specific branch
20:26:11 <andythenorth> and publishes that
20:26:48 <andythenorth> shall I just buy an Azure VM with Apache in it or something?
20:26:55 <andythenorth> this all seems super complex
20:27:19 <TrueBrain> yes, and getting a VM to put Apache on is not complex
20:27:22 <andythenorth> reminds me why I've seen all these devops CVs where the last 2 years experience is clicking AWS or Azure UI
20:27:33 <TrueBrain> but again, why not use GitHub releases for this?
20:27:38 <TrueBrain> why are you overcomplicating things? :)
20:28:22 * andythenorth reads about GH releases
20:28:50 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
20:29:09 <TrueBrain> for CompileFarm I just overwrite the same release every time
20:29:11 <TrueBrain> because ... why not
20:29:19 <andythenorth> and how do you tag it?
20:30:11 <andythenorth> how did you do that? :o
20:30:21 <andythenorth> the docs I've been reading all require tags
20:30:27 <andythenorth> annotated or lightweight
20:30:35 <TrueBrain> Why is this release on a random commit?
20:30:36 <TrueBrain> Although we publish a new windows-dependencies.zip often (at least once a week), the tag on which the release is made is never refreshed.
20:30:38 <andythenorth> I don't want to tag for a push build
20:30:40 <TrueBrain> it is .. right there :P
20:30:57 <TrueBrain> I just tagged a random commit once, which gave me a Release
20:31:03 <TrueBrain> and I am attaching a binary there every week since
20:31:13 <TrueBrain> (binaries don't have to be from that tag, as GH has no way to validate that, ofc)
20:31:46 <TrueBrain> call it a hack, call it "I don't give a fuck" :P
20:31:58 <TrueBrain> especially to just get it starting, it works fine
20:32:12 <TrueBrain> if you are happy with everything etc, and know what you want, we can always move stuff to either the CDN or via Musa
20:32:21 <andythenorth> to be clear, I'm not interested in releasing the grf yet
20:32:30 <andythenorth> I'm trying to web-publish the html docs
20:32:41 <TrueBrain> owh, sorry, that was not fully clear to me :P
20:32:47 <andythenorth> well I never said it :P
20:32:58 <TrueBrain> I assumed we were talking about binaries :)
20:33:04 <TrueBrain> how useful is the HTML without the binary?
20:33:08 <TrueBrain> feels like an anti-feature :)
20:33:21 <andythenorth> long-term it's silly
20:33:26 <andythenorth> but it's the more interesting problem to solve
20:33:41 <TrueBrain> euh ... DigitalOcean CDN is not good at hosting HTML files atm, so that is also not useful
20:33:49 <andythenorth> because the html has to be served from somewhere, with the MIME type configured so it doesn't download :P
20:34:07 <andythenorth> maybe this 'special branch for GH Pages' route works
20:34:09 <TrueBrain> is it a single HTML file?
20:34:15 <andythenorth> it's a static website
20:34:19 <TrueBrain> yeah, look into GH pages, sounds good
20:35:35 <andythenorth> then we can shut some coop VMS
20:35:35 <TrueBrain> so .. cmake for OSX still acts up ..
20:35:40 <TrueBrain> iconv is not working, and building fails :(
20:35:44 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z opened pull request #7310: Docs: table alignment when viewing README.md as a monospace textfile https://git.io/fhAH9
20:36:04 <TrueBrain> src/core/geometry_type.hpp:22:10: note: previous definition is here #define Rect OTTD_Rect
20:36:07 <TrueBrain> I have seen that before
20:36:10 <Eddi|zuHause> peter1138: i've used up all my spaces for at least the next 3 PRs :p
20:36:39 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain approved pull request #7310: Docs: table alignment when viewing README.md as a monospace textfile https://git.io/fhAHH
20:36:46 <peter1138> Ah, making it look nice without .md
20:37:58 <TrueBrain> GitHub is acting up all week
20:38:22 <Eddi|zuHause> ok, you deal with that :p
20:38:43 <TrueBrain> already done sir :)
20:39:06 <Samu> gonna experiment with "force_exact_match
20:39:25 <TrueBrain> ugh, this really is the moment I need access to a Mac :(
20:40:22 <Beerbelott> Eddi|zuHause: Relieved... rebasing my branch on top of trunk solved the regression pb... Dunno what happened these past few days on trunk :P
20:40:22 <Beerbelott> OK Now I'll dig into that OpenGFX loading trouble. That piqued my interest
20:40:49 <andythenorth> TrueBrain: I pay for macincloud? :P
20:41:14 <Eddi|zuHause> Beerbelott: -d misc=highnumber should tell about which directories it searches
20:41:54 <milek7> and linux don't care because filename is just opaque bytes
20:41:57 <Samu> 0x000000a503f4d260 "ludiai afterfix.7.7"
20:42:20 <peter1138> milek7, probably, yes.
20:42:24 <TrueBrain> milek7: lol .. nice find :)
20:42:35 <Beerbelott> Eddi|zuHause: Btw the regression on my PR was due to my wrong base on my branch... The push seems to have 'unstuck' it
20:42:38 <Eddi|zuHause> of course it's "Doc:" not "Docs:"
20:42:51 <TrueBrain> okay, this should fix OSX Iconv detection .. it is Iconv instead of ICONV .. lol
20:43:00 <TrueBrain> QuickTime not found, hmm
20:43:09 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z dismissed a review for pull request #7310: Docs: table alignment when viewing README.md as a monospace textfile https://git.io/fhAHH
20:43:10 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z updated pull request #7310: Docs: table alignment when viewing README.md as a monospace textfile https://git.io/fhAH9
20:43:14 <peter1138> Hmm, still 33 ms/t used by the PerformanceAccumulator.
20:43:53 <peter1138> Hmm, when you do a squash & merge, you can edit the commit title... what if I get the prefix wrong there? :p
20:44:02 <Eddi|zuHause> anyone can tell me in 3 easy steps how to run the commit-checker on push?
20:44:11 <TrueBrain> someone will drive to your house to smack you with a fish
20:44:14 <peter1138> Why not run it on commit?
20:44:33 <peter1138> Ok, grab the commit hooks and, er, put them in your commit hooks.
20:45:12 <milek7> hm, azure job timeout is 6 hours
20:45:16 <milek7> maybe i could start ssh on it and use it for development
20:45:17 <milek7> though it maybe qualifies as being bad neighbour ;p
20:45:41 <Eddi|zuHause> peter1138: where exactly does that go?
20:45:42 <Beerbelott> Eddi|zuHause: Again, running the game normally is not a pb, it's merely through bin/ai/regression/run.sh that the problem arises
20:45:59 <Beerbelott> I'll try to modify the script to make debugging options through
20:46:07 <peter1138> Well, the hooks directory part of it goes there.
20:46:18 <TrueBrain> milek7: I kinda like that MS is doing this; kinda want to keep them doing it ;)
20:46:40 <Beerbelott> Oh there are misc=9 already in there ;)
20:47:01 <milek7> even if i wanted to pay, they don't offer macos vm..
20:48:20 <TrueBrain> they don't host them; they use another party
20:49:31 <TrueBrain> 4 minutes .. "close" :P
20:49:42 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
20:49:52 <nielsm> I guess it is 10% over...
20:50:22 <nielsm> maybe I could have done without that stop midways to fill up on fuel
20:50:45 <Eddi|zuHause> well, now i have no idea what to test this with
20:51:19 <TrueBrain> seriously GitHub ... get your act together
20:52:42 <Eddi|zuHause> yay, this time CI passed :)
20:53:36 <milek7> they're using real macs for that, not vms
20:53:39 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain approved pull request #7310: Docs: table alignment when viewing README.md as a monospace textfile https://git.io/fhAHj
20:53:53 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain merged pull request #7310: Docs: table alignment when viewing README.md as a monospace textfile https://git.io/fhAH9
20:54:02 <TrueBrain> I got the hint Eddi|zuHause :P
20:55:30 <andythenorth> ok GH pages starts to make sense
20:55:49 <andythenorth> as usual no off-the-shelf solution works when you already have your own solutions in place
20:56:00 <andythenorth> it uses a weird orphaned branch
20:56:10 <andythenorth> which presumably AP will have to push the artifacts into
20:56:21 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
20:57:01 <andythenorth> it also talks about triggering a build on push to github
20:57:04 <andythenorth> not sure how that works
20:57:17 <andythenorth> I didn't know GH had CI built in
20:58:14 <TrueBrain> GitHub doesn't have a CI (yet)
20:58:28 <TrueBrain> it can only trigger stuff based on actions
20:59:13 <TrueBrain> andythenorth: from what I understand, GH Pages can publish any branch (even master) from a repository
20:59:35 <andythenorth> it's just built for a different problem to the one I have
20:59:44 <TrueBrain> andythenorth: what do you mean?
20:59:45 <andythenorth> but I think I can do bad things and it will work
20:59:52 <Eddi|zuHause> if it's in the results.txt, that means that is meant to happen :)
20:59:54 <Beerbelott> those are the results the current run are checked against during regression tests...
20:59:57 <TrueBrain> if you push your HTML files from AP to that branch, you should be fine
21:00:18 <TrueBrain> I am surprised how easy they made GH Pages ..
21:00:22 <TrueBrain> I should use them .. I guess
21:00:44 <andythenorth> pushing generated files to the repo has been a 'no-no' for so long
21:00:47 <andythenorth> it makes me twitch
21:00:58 <TrueBrain> but that is exactly what you want?
21:01:18 <TrueBrain> (well, it is not really the repo .. it is a branch on the repo which is a reflection of http)
21:01:23 <andythenorth> I've just been trained not to do it
21:01:43 <andythenorth> pretty much everyone I've ever collaborated with :)
21:01:47 <TrueBrain> make[2]: *** No rule to make target `/usr/local/lib/libpng.dylib', needed by `openttd'. Stop.
21:01:52 <TrueBrain> what is this bullshit now ..
21:02:03 <TrueBrain> hmm .. dynamic linking ..
21:02:06 <TrueBrain> that is not good for OSX
21:02:09 <TrueBrain> but okay, another issue
21:02:19 <andythenorth> is this the dynlibs that Apple fucked up silently?
21:02:22 <andythenorth> and then silently fixed?
21:02:30 <andythenorth> without bumping XCode versions?
21:02:47 <TrueBrain> hmm .. I wonder why that file is not there .. it was autodetected by CMake ..
21:02:54 <peter1138> I push generated files into repos... although they are deployment repos, so kinda necessary.
21:03:06 <Eddi|zuHause> Beerbelott: you likely want to look around the first line that is different between the (expected) result.txt and the (current) tmp.regression
21:03:17 <TrueBrain> andythenorth: that works too, what peter1138 says, just make a second repository where you publish your docs!
21:03:26 <peter1138> I should sue Tesco, these pitted prunes have pits :/
21:03:54 <Eddi|zuHause> peter1138: does that stuff work outside america?
21:05:36 <andythenorth> peter1138: you are no longer on my christmas card list
21:05:49 <peter1138> For using git for deployment?
21:05:53 <peter1138> Or for having prunes?
21:06:23 <TrueBrain> does it matter, really?
21:06:33 <peter1138> git for deployment is great.
21:07:42 <nielsm> remember, git is essentially blockchain for software development!
21:07:56 <peter1138> ^ funny, it's a patch to filter vehicles by cargo type, and says "the patch's name is self explanatory" and the patch's name...? group-name-patch.
21:08:57 <TrueBrain> andythenorth: where is libpng.dylib located on your machine?
21:10:16 *** Thedarkb-T60 has joined #openttd
21:10:22 *** andythenorth has joined #openttd
21:10:22 <TrueBrain> owh, I see what is going wrong here .. interesting
21:10:45 <TrueBrain> ha, yes .. we cause this ourself
21:10:48 <TrueBrain> okay .. that is wrong :D
21:11:42 <TrueBrain> it indeed are rms causing this :)
21:11:51 <TrueBrain> in the old config.lib, it worked
21:11:59 <TrueBrain> now CMake is like: brew told me it should be here!
21:12:38 <andythenorth> but I already have master
21:13:02 <andythenorth> I think these aren't the docs I need
21:13:23 <TrueBrain> in the Settings of your project
21:13:28 <TrueBrain> you can select which branch to use for GH Pages
21:13:36 <glx> well if we remove the dylib we may need to enable static linking :)
21:13:54 <andythenorth> yes I found that
21:14:10 <andythenorth> but I created the gh-pages branch incorrectly, so I'm doing it again
21:14:19 <andythenorth> why is the web so fucking awful now?
21:14:46 <andythenorth> I go on a page about git, and I get cookie warning, GDPR warning, push notification, and they want me to get an app
21:14:51 <glx> btw mingw will need static too, currently it compiles but it's not very usable outside mingw terminal
21:16:20 <nielsm> it's a better train simulator
21:16:27 <TrueBrain> glx: yup, something like that indeed
21:17:57 <peter1138> nielsm, is the timer fixed?
21:18:05 <TrueBrain> weird; in static mode, pkg-config makes mistakes :(
21:19:33 <nielsm> peter1138: the time allotted depends on the distance, but going over only means you don't get the bonus
21:19:42 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
21:19:49 <peter1138> nielsm, right but is it realtime? so 20 minutes is actually 20 minutes.
21:20:08 <nielsm> it's not like ETS2 scale
21:20:13 <andythenorth> orphan branches eh?
21:20:19 <peter1138> In the demo everything was about 3x too fast.
21:23:52 <andythenorth> for an orphan branch, can I leave my old gitignore in place?
21:23:56 <andythenorth> the instructions say not
21:24:21 <andythenorth> oh fuck it, I just do it and see what goes wrong
21:24:26 <andythenorth> nobody learns anything by talking
21:24:28 <TrueBrain> funny .. it cnanot find iconv in static mode ... this is weird :P
21:24:44 <TrueBrain> owh, all runs fail .. haha
21:26:25 <andythenorth> so /docs is in my .gitignore of course
21:26:31 <andythenorth> but it can't be for this branch
21:28:45 <andythenorth> they're not published from AP
21:28:50 <andythenorth> that's a local commit
21:29:29 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
21:29:38 <TrueBrain> but you have GH Pages to work!
21:30:17 <andythenorth> the magic is good
21:30:28 <andythenorth> as usual, figuring out which of the docs instructions I need was the hard bit
21:30:53 <andythenorth> I should remember that for my customers :P
21:32:10 <andythenorth> this needs to be on the coop account somehow
21:32:16 <andythenorth> that's probably fine
21:33:07 <andythenorth> right now to teach azure
21:33:09 <milek7> do macos have smartscreen-like annoying popups?
21:33:29 <TrueBrain> okay, OSX still fails; possibly some CFLAGS that is causing issues ... something for another day
21:35:17 <andythenorth> I don't know exactly what those are, but there are annoying popups
21:38:07 <Samu> i'm trying to use "force_exact_match bool, instead of -2
21:38:44 <milek7> i have private key on windows virtual smartcard, which is on kvm swtpm
21:38:54 <milek7> and now how to extract key from that? ;D
21:39:30 <TrueBrain> if it is any decent, you can't :P
21:40:21 <milek7> swtpm is software only so it should be possible
21:40:33 <Samu> alright, i got somewhere
21:45:57 *** supermop_Home has joined #openttd
21:46:34 <Beerbelott> Eddi|zuHause: Problems indeed arise because of -c ai/regression/regression.cfg
21:47:09 <Beerbelott> stripping this parameter from regression scripts makes a failed regression test due to no detection of OpenGFX pass
21:47:24 <Beerbelott> so sth messes up w/ dir detection, there
21:49:02 * andythenorth has infosec horrors
21:49:13 <andythenorth> giving unknown extensions write access to my repos
21:51:29 *** supermop_Home_ has quit IRC
21:54:00 <andythenorth> "The current operating system is not capable of running this task. That typically means the task was written for Windows only. For example, written for Windows Desktop PowerShell."
21:54:18 <andythenorth> can't run powershell on ubuntu then :P
21:58:12 <andythenorth> ok I might need actual help :)
21:58:16 <andythenorth> from actual programmer
21:58:34 <Eddi|zuHause> where would you get one of those from?
21:59:06 <andythenorth> I think one just volunteered?
21:59:51 <andythenorth> TrueBrain: so if I make a Service Connection to github, I can auth that to push back to my repo?
22:00:02 * andythenorth joining final dots, hopefully
22:00:13 <TrueBrain> andythenorth: I think so, yes
22:05:54 <andythenorth> am I now the in-house AP expert? o_O
22:09:33 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
22:10:36 <TrueBrain> I forgot, OSX cannot handle '-static' at ALL :P
22:11:18 <TrueBrain> the connection between AzurePipelines and GitHub is a lot better now .. it used to be < 100mbit/s .. it is now ~300mbit/s
22:11:20 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on pull request #7289: Add: Configurable ship curve penalties (YAPF) https://git.io/fhA7o
22:15:20 <TrueBrain> okay .. this doesn't work at all now; lol
22:15:38 <Eddi|zuHause> sounds normal :p
22:15:40 <milek7> eh, why .so couldn't be used for static linking?
22:15:51 <TrueBrain> because ... .so by definition is dynamic?
22:16:55 <glx> 'fontconfig' depends on 'm', of which no static variant was found <-- of course ;)
22:16:59 <milek7> it contains code and symbol names, what more is necessary?
22:17:15 <peter1138> Hmm, the problem with itch-scratching PRs is nobody else cares enough to look at them.
22:17:31 * peter1138 looks at issues instead.
22:17:48 <TrueBrain> milek7: I suggest google as your friend :)
22:18:25 <TrueBrain> glx: yeah .. PkgConfig is really terrible with STATIC, so my current code can contain some issues :)
22:18:43 <milek7> it didn't helped, besides such handwaving as '.a is for static libraries and .so is dynamic'
22:18:49 <Eddi|zuHause> well, on a theoretical level, you could ask why the .so fileformat wasn't prepared in a way that you could strip it of the dynamic bits and link it like you would an .a file
22:19:05 <glx> well -lm is not a real lib IIRC
22:19:13 <TrueBrain> glx: OSX is weird .. at soon as I requested static, the default CMake files failed
22:19:26 <TrueBrain> glx: exactly why I mention that it is my own code ;)
22:19:36 <TrueBrain> hmm .. I want to see the cmake file on OSX
22:19:42 <TrueBrain> maybe that gives a clue why it is acting like this
22:20:01 <nielsm> but, does cmake then know how to collect and re-install_name dylibs so they work in an .app package?
22:20:17 <nielsm> if it can do that, that would be just as good wouldn't it?
22:20:21 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN approved pull request #7147: Change: Synchronize randomness in vehicle introduction… https://git.io/fhA7D
22:20:25 <TrueBrain> hmm .. you are not wrong
22:20:33 <TrueBrain> we just never investigated that, as it was impossible manually :D
22:21:19 <peter1138> "Change: Synchronize randomness in vehicle introduction so all vehicles introduced simultaneously will stay at the same date (#7147)"
22:21:31 <peter1138> That ... is a long winded and nonsensical commit message.
22:21:45 <Eddi|zuHause> i accept alternative wordings
22:22:16 <nielsm> keep same-intro-date vehicles on same intro date
22:22:44 <peter1138> No need to change anything, I can reword the commit message.
22:22:56 <Eddi|zuHause> nielsm: i don't think that makes it any clearer
22:23:00 <peter1138> Just... it's almost Samu-like ;p
22:24:22 <Eddi|zuHause> keep in mind i'm not a native speaker.
22:24:31 <peter1138> "Change: Synchronize introduction date randomness across vehicles with the same base date" ?
22:24:38 <milek7> Eddi|zuHause: but is there something missing in .so elf? or just nobody thought to implement it in linker
22:24:58 <peter1138> *base introduction date
22:24:59 <Eddi|zuHause> milek7: how should i know?
22:25:21 <Eddi|zuHause> peter1138: maybe add a bit about reliability curve?
22:25:51 <TrueBrain> glx: that cannot work, you are missing quiet a few instances :)
22:25:59 <peter1138> Change: Synchronize introduction date and reliability randomness across vehicles with the same base introduction date (#7147)
22:25:59 <TrueBrain> but if something is always Release, why not just add it to windows-build.yml?
22:26:16 <peter1138> Still long but works I think.
22:26:32 <glx> because testing will need it too
22:26:44 <TrueBrain> which you also set on Release
22:27:16 <TrueBrain> a variable that is a constant is a bit annoying in Azure Pipelines
22:27:36 <glx> I set a job variable and pass it to the VSBuild tasks
22:28:10 <Eddi|zuHause> peter1138: sounds fine. but do we really want to rush this in before RC1 release? ;)
22:28:25 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN merged pull request #7147: Change: Synchronize randomness in vehicle introduction… https://git.io/fhDkW
22:28:28 <TrueBrain> meh, why is static always such a pain in the ass ..
22:29:03 <glx> but in azure pipelines doc they suggest "Tip: Declare a build variable such as BuildConfiguration on the Variables tab (selecting Allow at Queue Time) and reference it here as $(BuildConfiguration). This way you can modify the platform when you queue the build and enable building multiple configurations."
22:29:06 <supermop_Home> andythenorth does alcohol increase hotel production?
22:29:26 <glx> which seems to be a better option
22:29:31 <TrueBrain> glx: my suggestion to you is to not use a variable; take it as it is, or do it another way, I don't care ;) Debug vs release is the least of my issue atm :)
22:29:36 <supermop_Home> trying to scrape together some $$ on this coop server
22:29:45 <andythenorth> FML, how to delete a pipeline in AP?
22:29:46 <TrueBrain> your current commit only is not working, as you forgot a few things .. look at other parameters to find all the instances you need to create
22:29:50 <andythenorth> without deleting whole project?
22:30:03 <nielsm> well could be the hotel guests ended up producing more guests to take home after consuming alcohol...
22:30:04 <peter1138> andythenorth, it's not hg...
22:30:13 <TrueBrain> andythenorth: just ... click on delete?
22:30:22 <TrueBrain> (click on the pipeline, click on the dots, click on delete)
22:31:36 <TrueBrain> okay, CMake has some bundle code, so possibly it can indeed copy the dylibs
22:31:40 <TrueBrain> would be fancy, I guess
22:31:55 <TrueBrain> still I am a bit puzzled why it failed to find a libz that is static
22:32:01 <glx> oh you mean all compile stuff, not only CI
22:32:29 <nielsm> osx ships a libz.dylib in /usr/lib which you can safely link
22:32:42 <nielsm> (or maybe it's in /lib)
22:32:48 <TrueBrain> glx: check all usages of VcpkgTargetTriplet
22:32:56 <andythenorth> "Bash exited with code '2'."
22:33:01 <TrueBrain> nielsm: yeah; I was wondering about the static variant
22:33:04 <TrueBrain> but I don't think it exists
22:33:10 <TrueBrain> as I think we "cheat" a bit with the current binaries
22:33:21 <TrueBrain> we remove the dynamic version of the libs, so it picks the static of the ones we added
22:33:26 <TrueBrain> but picks the dynamic of everything else
22:33:35 <peter1138> If it's shipped with osx, can't we use it?
22:33:38 <TrueBrain> so we are a bit mixed :D
22:33:56 <peter1138> Static is bad for security.
22:34:10 <TrueBrain> that is a weird and untrue statement :)
22:34:26 <TrueBrain> I have used static binaries often to patch security issues
22:34:29 <TrueBrain> as the OS was too slow
22:35:27 <LordAro> depends on the system
22:35:42 <TrueBrain> not really ;) Depends on which maintainer is quicker
22:35:49 *** LordAro sets mode: -o LordAro
22:35:53 *** DorpsGek sets mode: +o LordAro
22:36:04 *** DorpsGek sets mode: -o LordAro
22:36:10 <peter1138> That is a rare case.
22:36:31 <peter1138> In general, our users who are still using 1.8.0 would be missing any security updates since then.
22:36:46 <LordAro> lots of static binaries means you have to find and recompile/fix every single one of them
22:36:48 <Eddi|zuHause> we've had people come in here using 5 year old game versions
22:36:57 <peter1138> Of course, this is hardly security critical software, but the point still stands.
22:37:03 <LordAro> dynamic library -> replace the library, done
22:37:07 <peter1138> If we can use system provided dynamic libs, we probably should.
22:37:18 <TrueBrain> LordAro: you weren't around when OpenSSL "bumped" their version? :P
22:37:35 <TrueBrain> peter1138: we are talking about either shiping the dylibs, or building it static
22:37:39 <TrueBrain> so not much difference in infosec terms
22:37:50 <peter1138> We shouldn't ship a dynamic lib if it's provided by the system.
22:38:16 <LordAro> but yes, of course this is dependent on library maintainers "doing it properly" as well
22:38:32 <LordAro> but these aren't windows systems, they have sane library management
22:38:38 <TrueBrain> it is all so fluffy, there is not one true statment :D
22:38:50 <peter1138> One is more truer :p
22:38:59 <andythenorth> ow I don't understand AP :(
22:39:06 <andythenorth> this makes me full 100% samu
22:39:32 <peter1138> You should spend hours debugging something that we've already explained...
22:39:33 <LordAro> general case vs real life, i think
22:39:33 <TrueBrain> but okay, the dutch saying: "de beste stuurlui staan aan wal", applies here :)
22:40:04 <TrueBrain> okay, config.lib makes our OSX binary some hybrid .. because we modify the build system sufficiently
22:40:10 <andythenorth> there's some way to auth a pipeline to push back to my repo, but FML I can't make it work
22:40:14 <TrueBrain> did not realise it yet like that
22:40:21 <andythenorth> also I'm in wife trouble :P
22:40:28 <LordAro> "the best drivers are on shore"
22:40:31 <TrueBrain> go fix that first andythenorth :)
22:40:46 <andythenorth> the things I didn't do are already not done
22:40:59 <Samu> I'm doing changes inside ai_sl.cpp, to make it use force_exact_match variable
22:41:00 <LordAro> switched my keyboard layout to en_US
22:41:13 <TrueBrain> andythenorth: you need to have a Service Connection to GitHub with write access to your repo
22:41:14 <Samu> would help me eliminate everything from ai_scanner.cpp
22:41:20 <TrueBrain> and I think if you checkout with that connection, it can also push
22:41:37 <peter1138> LordAro, keeps doing that to me. You need to uninstall it every time it gets reinstalled :/
22:41:54 <glx> I set a job variable and use them in the steps
22:41:58 <LordAro> yeah, except now i don't have an option to uninstall it :(
22:42:20 <TrueBrain> now glx is acting like Samu :D
22:42:23 <TrueBrain> glx: CONTEXT PLZ :D
22:42:51 <glx> I don't see what's wrong in my commit
22:43:03 <TrueBrain> ah; did you follow my advise? and grep that other variable?
22:43:10 <TrueBrain> it will tell you the instances you missed just fine :)
22:44:29 <peter1138> Can I review my own PRs?
22:44:53 <peter1138> I want to be a rogue element :p
22:45:05 <glx> and I can't found any VcpkgTargetTriplet in other files
22:45:13 <andythenorth> 'azure push to github' is a crap thing to search for
22:45:18 <Eddi|zuHause> peter1138: make another account and give it maintainer rights? :p
22:45:21 <andythenorth> many many results, and not what I want
22:45:46 <peter1138> Eddi|zuHause, easier, make another account, and approve all its PRs :p
22:46:15 <glx> release.yml is not converted to cmake yet
22:46:16 <peter1138> No, Steam, X-Plane 11 isn't running.
22:46:35 <Eddi|zuHause> nothing a killall -9 won't fix :p
22:47:09 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
22:47:33 <TrueBrain> you are missing an entry here
22:47:56 <TrueBrain> you are missing an entry here
22:48:38 <milek7> generic hacky way: place ssh private key with write permissions on secret url, use that url as secret variable in AP, download it and use in git
22:48:57 <TrueBrain> milek7: that is not a hacky way, that is how dataleaks are created
22:49:17 <glx> but release.yml needs more changes too
22:49:24 <milek7> why? where it would leak?
22:49:35 <TrueBrain> milek7: that "secret" URL, ofc
22:49:43 <glx> I can include them in the 'Release' stuff
22:50:05 <TrueBrain> milek7: security by obscurity never worked really :)
22:50:18 <TrueBrain> (more generic, copying a private key EVER is always a bad idea :P)
22:50:36 <milek7> random file name on directory with disabled autoindex is undiscoverable
22:50:49 <TrueBrain> yeah .. and there is no way that variable is going to leak :P
22:51:03 <LordAro> mm, it's fine... until it's not
22:51:17 <LordAro> and you have no way of knowing when it's not
22:51:39 <TrueBrain> (when you find porn on your github repos :P)
22:51:50 <TrueBrain> yes, the worst thing I could think of, was porn
22:52:24 <TrueBrain> "ld: library not found for -llzo2"
22:52:28 <TrueBrain> what a lame ass response is that
22:53:01 <TrueBrain> pkg-config on OSX is bad
22:56:17 <DorpsGek_II> [OpenTTD/OpenTTD] SamuXarick updated pull request #7193: Fix #6468: Load correct version of AI as specified during the time of its save. https://git.io/fhHI1
22:58:20 <DorpsGek_II> [OpenTTD/OpenTTD] SamuXarick updated pull request #7193: Fix #6468: Load correct version of AI as specified during the time of its save. https://git.io/fhHI1
22:58:46 *** andythenorth has joined #openttd
22:58:46 <peter1138> I don't have any saves with AIs, let alone any in a condition to be able to test this.
22:58:49 <Samu> damn visual studio style getting in the way
22:59:04 *** andythenorth has left #openttd
22:59:51 <DorpsGek_II> [OpenTTD/OpenTTD] SamuXarick updated pull request #7193: Fix #6468: Load correct version of AI as specified during the time of its save. https://git.io/fhHI1
23:00:40 <Samu> about src/script/script_config.cpp I'm still unsure if I can remove it
23:01:25 <peter1138> Azure is really not having much luck today :/
23:04:59 <LordAro> Samu: make a minimal example
23:05:08 <LordAro> should be pretty trivial
23:05:17 <LordAro> doesn't have to be based on any existing AI
23:06:14 <Samu> gonna test with ommissing v7 on purpose
23:06:20 <TrueBrain> glx: I think so too, yes
23:08:01 <DorpsGek_II> [OpenTTD/OpenTTD] glx22 updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
23:08:58 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
23:09:04 <TrueBrain> okay, this might just fix OSX ... *fingers crossed*
23:10:21 <TrueBrain> but GitHub API is being a bitttcchhhh
23:11:15 <Samu> just tested, it's doing wrong t.t
23:11:34 <TrueBrain> so we either have to ship dylibs with OSX, or we have to find a way to be selective what we make static .. or find a static zlib ..
23:11:44 <TrueBrain> (dynamic, hybrid, static \o/)
23:13:11 <nielsm> what's wrong with using apple's supplied libz.dylib?
23:13:40 <TrueBrain> nielsm: I am fine with the libz, but again, that is not the issue :D
23:13:45 <glx> but that was a typical lib order failure
23:13:46 <TrueBrain> we have 4 libraries we download via brew
23:13:54 <TrueBrain> we can safely assume users don't have them
23:14:05 <TrueBrain> so we either have to ship them to the user as dylib, or compile them static
23:14:50 <TrueBrain> we used to do that, by removing the dylib, and magically the compiler pick the static version
23:15:02 <TrueBrain> CMake ruins that, as it validates that the dylib is there, if it expect to find it
23:15:11 <DorpsGek_II> [OpenTTD/OpenTTD] SamuXarick updated pull request #7193: Fix #6468: Load correct version of AI as specified during the time of its save. https://git.io/fhHI1
23:15:24 <TrueBrain> (in other words, we used to be hybrid: dylibs for those that are not fetched via brew, static for those via brew)
23:15:40 <TrueBrain> so my first attempt was to do everything static .. but zlib doesn't have a static version on OSX
23:15:53 <TrueBrain> nielsm: what does that do?
23:16:23 <nielsm> it enumerates all dylibs linked, recursively, copies into an .app package, and adjusts install_names for them to search for those libs inside the package
23:16:27 <TrueBrain> okay, OSX now works with dylibs .. that ati least is something :D
23:16:36 <TrueBrain> nielsm: yeah, that works indeed :)
23:16:47 <TrueBrain> it seems CMake has something similar
23:16:51 <TrueBrain> but that means we are bundling the dylibs
23:16:57 <TrueBrain> so that will be fun to see how that goes :)
23:17:08 <TrueBrain> (also the bundles will most likely be larger, but who cares)
23:17:23 <TrueBrain> I guess the main issue here is nielsm, that we have a similar issue for Windows (via mingw) and linux (generic builds)
23:17:27 <TrueBrain> suggestions for those?
23:17:56 <Beerbelott> Eddi|zuHause, peter1138: the error relative to missing graphics set in the regression script comes from src/bootstrap_gui.cpp on line 220
23:18:09 <Beerbelott> if (BlitterFactory::GetCurrentBlitter()->GetScreenDepth() == 0) goto failure;
23:18:24 <Beerbelott> The error only appears when the regression cfg file is being used
23:18:27 <TrueBrain> glx: but compiling X11 staticly is not the best approach on Linux ;)
23:18:35 <Beerbelott> Does that ring any bell?
23:18:55 <Eddi|zuHause> Beerbelott: that just says "if headless, go into error instead of asking the user"
23:19:14 <Eddi|zuHause> Beerbelott: the failure to detect the graphics is earlier
23:19:18 <Beerbelott> But by no loading the special config file, it works like a charm
23:19:24 <glx> for windows it's easier as all searched libs are all non windows
23:19:54 <Eddi|zuHause> Beerbelott: yes, but that has nothing to do with that line
23:20:18 <TrueBrain> so that leaves Linux in the middle :P
23:20:40 <Eddi|zuHause> Beerbelott: that line is already deep in the "we found no set, go into panic mode" path
23:21:39 <TrueBrain> okay, all targets now fail on regression, as they should \o/
23:21:46 <TrueBrain> far from done, but at least another step
23:22:06 <Samu> arf... i really need v8 or v6
23:22:10 <TrueBrain> nielsm: yeah, that osx-fix-libs does what I was considering, assuming /usr/local is "bad" .. still a bit hackish, but I guess it will have to do
23:22:48 <TrueBrain> nielsm: would there be a difference in making those libraries static vs bundling the dylibs, you think?
23:22:52 <nielsm> TrueBrain, that script was written with the assumption that you do not want to link in libs from homebrew or macports or similar (because they may be "weird")
23:23:13 <nielsm> for osx, no, dylibs in the app package is standard operating procedure
23:23:36 <TrueBrain> well, we don't really package mingw, so I guess strictly seen we don't have to fix that
23:23:42 <TrueBrain> so no static for now; will revisit that some day :P
23:23:54 <nielsm> I believe Xcode's own build even does all that install_name magic
23:24:30 <Samu> so i cloned v9 and changed info.nut to say this is v8
23:24:53 <TrueBrain> nielsm: good to know :)
23:25:00 <LordAro> Samu: might want to change some of the debug messages so you can more easily tell which version is which
23:25:34 <Samu> need to start v8, save, then delete v8: expected result is to load v9 without a DEBUG warning
23:26:09 <TrueBrain> glx: as said, CMake supports bundling ;)
23:26:46 <TrueBrain> okay .. next on the list ... the horror: ICU
23:29:26 <milek7> it should be possible to integrate selected elf dependecies into elf itself: copy library code section, perform relocations to library placed now in application address space, and copy .so relocation table to main elf with offset to where library code was copied
23:38:33 <Samu> ok it loaded v9, no warning, fine so far, but then, the name still remained with .8 attached in the end
23:39:09 <Samu> i still need that code in script_config.cpp
23:40:55 <TrueBrain> lol .. ICU "just works" .. that was unexpected :D
23:41:38 <TrueBrain> had to rename LAYOUT to lx and SORT to i18n
23:41:41 <TrueBrain> but .. that was all :)
23:41:57 <TrueBrain> libpthread.so.0: error adding symbols: DSO missing from command line
23:42:56 <milek7> find_package(Threads REQUIRED) missing?
23:43:14 <TrueBrain> no .. one of the pkg-config libraries doesn't announce it
23:43:18 <TrueBrain> pkg-config is horrible :P
23:43:35 <TrueBrain> okay, a bit more verbose: it is compiling fine when I use find_package for everything
23:43:42 <TrueBrain> but it broke when using pkg-config for everything
23:43:54 <TrueBrain> (and normally pkg-config is a fallback in case the cmake Find is not installed)
23:44:51 <DorpsGek_II> [OpenTTD/OpenTTD] SamuXarick updated pull request #7193: Fix #6468: Load correct version of AI as specified during the time of its save. https://git.io/fhHI1
23:45:46 <TrueBrain> well, I guess your suggestion works fine to fix that milek7 :P
23:45:53 <TrueBrain> but it is stupid pkg-config is this bad :(
23:46:12 <peter1138> Should use autotools/autoconf/automake/etc instead ;)
23:46:27 <TrueBrain> you do know those tools use pkg-config too, right?
23:47:03 <TrueBrain> pkg-config was going to solve all the library detection problems .. instead it just created an other way of things to go wrong :P
23:49:17 <Samu> now cleaning up into just 1 commit
23:49:43 *** Thedarkb-X40 has joined #openttd
23:50:06 <TrueBrain> still makes me laugh, those names
23:50:15 <TrueBrain> Eddi|zuHause: exactly that, yes
23:50:26 <TrueBrain> I like that when ever you say something like that, always someone finds the XKCD link for you :P
23:51:17 <DorpsGek_II> [OpenTTD/OpenTTD] SamuXarick updated pull request #7193: Fix #6468: Load correct version of AI as specified during the time of its save. https://git.io/fhHI1
23:51:55 <TrueBrain> I have layout/LETypes.h in icu-le-hb
23:53:17 <LordAro> TrueBrain: i wouldn't expect icu-le-hb at all, unless you're fixing 6922
23:53:32 <TrueBrain> no clue how it got on my system :)
23:53:38 <TrueBrain> what is the normal folder for LETypes.h?
23:57:25 <TrueBrain> nevermind; manually detection is too complex
23:57:33 <TrueBrain> so either CMake or pkg-config required for ICU support
23:58:21 <LordAro> TrueBrain: there is an icu-config binary, if that's relevant
23:59:06 <TrueBrain> I am hoping we dont have to support any NNN-config except pkg-config
23:59:09 <TrueBrain> but time will tell :D
continue to next day ⏵
