Gideros 2022.1 Released

rrraptor

@hgy29 Is it possible to turn off specific linter checks?

Or maybe fix this one?)

P.S.

I accidentally clicked the "Dismiss" button and now I don't see the topic in the general list, how can I fix it?)) I didn't find any settings in my profile.

hgy29

Currently some of the checks are already disabled by default, and you can disable all purple checks by unselecting 'type checking'. My plan is to give more information to the linter/typechecker in future Gideros versions, currently it only parse one file at a time, and it doesn't know about Gideros API.

keszegh

@hgy29, in midi plugin is there a way to prevent my app crashing from
"MidiInWinMM: message queue limit reached!!" when too many messages arrive? (which happens easily when a pot is turned on a midi device)
ty

EDIT: is there a page where one can check the available lua functions for the midi plugin?

EDIT: exposing setQueueSizeLimit( unsigned int queueSize ) could help perhaps, but all in all i'd need some way to stop the app from crashing whatever the queue size.

hito9

Can anyone else test-confirm these examples below on a solid Windows 10 installation?
Curves
3D Instancing
Particles
Bird Animation 3D
MapClown

For Windows 10 on Virtualbox, The Player has upside-down
rendering. Besides please notice "click to walk" text mirroring.
At the same time mouse touch points and canvas rendering are reversed.
Exported 3d Animation example screencast link(lossy gif)

https://imgbox_dot_com/WSRkfI5l

Here is the steps to produce another repetitive bug(probably destructor issue).
Firstly run the "3D Collision" example. Please be sure while the boxes were
in constant multiple contacts(movement) and launch the "Normal Mapping" example.
The Player crashes. I hope it helps

MoKaLux

hito9 said:

Can anyone else test-confirm these examples below on a solid Windows 10 installation?

I am running windows 10 64bits:
Curves OK
3D Instancing OK
Particles OK
Bird Animation 3D --> almost OK the right eye isn't seeing any birds
MapClown OK

For Windows 10 on Virtualbox, The Player has upside-down
rendering. Besides please notice "click to walk" text mirroring.
At the same time mouse touch points and canvas rendering are reversed.
Seems ok for me


Exported 3d Animation example screencast link(lossy gif)
Exported 3d Animation exported to .exe crashes right away for me

Here is the steps to produce another repetitive bug(probably destructor issue).
Firstly run the "3D Collision" example. Please be sure while the boxes were
in constant multiple contacts(movement) and launch the "Normal Mapping" example.
The Player crashes.
Can confirm that too on my machine, but it is a known bug and it is being taken care of.

I hope it helps
That certainly helps

New gideros luau looks awesome

MoKaLux

@hito9 be sure to have the same settings in the project (right click your project name) and the player:

hito9

MoKaLux said:

@hito9 be sure to have the same settings in the project (right click your project name) and the player:

@MoKaLux Thank you for testing awesome new Gideros I am glad to see your Gideros is ok on Windows 10. I had tested with the default example settings for my feedbacks. It seems VirtualBox graphics driver is the source of problems except your exported project and Bird Animation 3D example(the right eye issue). Normally it looks like this
giderosmobile.com/examples#VR/Bird%20Animation%203D
Also the specific error is triggerring when closing some exported examples. ParticlesClown example is overscaling all particles after 300 frames for my 2021.11 Linux build and wine 2022.1.1 Luau version.


Unlike this online example
giderosmobile.com/examples#Graphics/ParticlesClown

MoKaLux said:

Can confirm that too on my machine, but it is a known bug and it is being taken care of.

Yes i know no rush, Our goal is the same. Already Gideros 2021.11 works great and Gideros Luau version is very promising.

MoKaLux

yes can confirm the clown particles sample and (at)hito9 I am no gideros developer yet!?

elves

Tried to install 2022.1.1 on my M1 Mac.

Previous one was working OK, so I might have to go back to that.

The studio runs but the player crashes. The detailed crash log says its not translating the x86 code.

SinisterSoft

I'll take a look at those samples.

hgy29

elves said:

The studio runs but the player crashes. The detailed crash log says its not translating the x86 code.

Can you post or otherwise send me the detailed crash log ? I don’t have a M1 to test

elves

It's working now! And I can get more warnings, too!

Initial order of installation was: remove 2021.11 and installed 2022.1.1 - that crashed.
So while I waited for a response I re-installed 2021.11
When you asked for the crash log I couldn't find it.

This time I just installed 2022.1.1 over my existing 2021.11 and it worked. And it really says it is 2022.1.1 player running.

I will go around the cycle one more time to see if the first order repeats it's failure.

No, clean install of 2022.1.1 works fine. There's a slight flickering on the Outline panel in Studio and the player won't start from Studio. When I start player separately, it all works.

I might try compiling my own player and see what happens.

Thanks for your patience with me.

rrraptor

Ive noticed another function:

typeof(values)

which should call "__type" metamethod and return the result of it, but it does not work for some reason :?

Code:

type Point = { x: number, y: number } local point : Point = {x = 10, y = 20}   Foo = {}   function Foo:new() local o = {} setmetatable(o, self) self.__index = self return o end function Foo:__type() return "Foo" end local foo = Foo:new()   Bar = Core.class()   function Bar:__type() return "Bar" end local bar = Bar.new()   print(typeof(foo)) --> "table" Not OK print(typeof(bar)) --> "table" Not OK print(typeof(Shader)) --> "table" Not OK?   print(typeof(point)) --> "table" OK? print(typeof(123)) --> "number" OK print(typeof("A")) --> "string" OK print(typeof(false)) --> "boolean" OK print(typeof(print)) --> "function" OK

"typeof" function
https://github.com/Roblox/luau/blob/8fe95c99630063ead551303d3ecb6888baeb911f/VM/src/ltm.cpp#L117

rrraptor

@hgy29 with new ternary operator code folding breaks.

hgy29

rrraptor said:

@hgy29 with new ternary operator code folding breaks.

Yes, I can understand that. I wonder if I'll be able to patch the lua lexer to cope with this.

rrraptor

hgy29 said:

rrraptor said:

@hgy29 with new ternary operator code folding breaks.

Yes, I can understand that. I wonder if I'll be able to patch the lua lexer to cope with this.

mb vscode extension can help you?)

https://github.com/Dekkonot/vscode-luau

hgy29

I am afraid it doesn't help; current lexer just detects 'if' and expect a 'end' afterwise to close the folding level. ternaries do have the 'if', but no 'end'

rrraptor

hgy29 said:

I am afraid it doesn't help; current lexer just detects 'if' and expect a 'end' afterwise to close the folding level. ternaries do have the 'if', but no 'end'

Then maybe take a look on Luau source code?

https://github.com/Roblox/luau/blob/8fe95c99630063ead551303d3ecb6888baeb911f/Ast/src/Parser.cpp#L2265

hgy29

Sure I know, and we actually parse the code for the linter, but doing it as part as lexilla is somewhat complex. I'll try to use heuristics in lexilla to overcome this

hgy29

Ok, I think I have something that should works: I assume ternary operators always follow an operator symbol, or something that expect an expression after. If the 'if' is at a place where an expression is expected, then we know it is a ternary, and ended by the 'else'

rrraptor

One more report
Seems like new scintilla editor does not work properly with custom styles.

MoKaLux

hopefully bugs will be fixed, big bugs then small ones

hgy29

Could it more a bug with QT6 than the editor ? Or maybe the way the new editor is wrapped in QT differs from the old.

rrraptor

hgy29 said:

Could it more a bug with QT6 than the editor ? Or maybe the way the new editor is wrapped in QT differs from the old.

I can't say because I don't know where to look and what to look for... If you can tell me how to build the studio, then I can try to find out what's wrong.

If I understand correctly, I need to build the project in "ui/ui.pro", but I can't. I have compiled "lexilla" and "scintilla", but when I try to build the main project, I get an error that the header file was not found ("#include "ScintillaEdit/ScintillaEdit.h" ")
although both libraries are in the folders: "lexilla/src/debug/Lexilla5.dll" and "scintilla/qt/ScintillaEdit/debug/ScintillaEdit5.dll" (I also tried the release version).
For some reason QT can't find the library?

hgy29

Yes, that's because I used to download lexilla and scintilla from official zip files, and build them seprately. In that case the header path would change so I decided to copy headers in QT installation. But later I needed to modify lexilla and scintilla, so I set up a git repo for them, but I didn't want to change all my #include to point to the new paths, so I decided to leave them in QT.

To build scintilla, you can use

make -f scripts/Makefile.gid qscintilla

.
Same for lexilla and debug versions of the libs: qlexilla, qscintilla.debug, qlexilla.debug

qscintilla & qlexilla targets will copy the headers where Gideros expects them

rrraptor

Or I can change this line?)
https://github.com/gideros/gideros/blob/6f4234c8002505776d9f1a841d14f6f2e84a37e1/ui/ui.pro#L263

hgy29

No, you would want to change the INCLUDE_PATH instead

hgy29

As promised last week, I have built and uploaded Gideros 2022.1.2, with its set of fixes and improvements over 2022.1.1

keszegh

thanks, now setFullSceen does not crash anymore. very good.

can we have an Application:getWindowSize? (i want to get it so that i can set the same when i quit fullscreen as it seems to reset the size to the initial window size).

btw shouldn't it be the default behaviour of setFullScreen(false) to recover the last screen size before going to fullscreen?

hgy29

keszegh said:

can we have an Application:getWindowSize? (i want to get it so that i can set the same when i quit fullscreen as it seems to reset the size to the initial window size).

btw shouldn't it be the default behaviour of setFullScreen(false) to recover the last screen size before going to fullscreen?

Yes that's what should happen in theory, but for some reason it doesn't work. If you want to do it at lua level, maybe listening to APPLICATION_RESIZE events could help ?