Replies: 22 comments 64 replies
-
|
@raysan5 what can SDL do what GLFW can't? |
Beta Was this translation helpful? Give feedback.
-
|
@raysan5 there are couple of open-source programs which are also used by existing open-source projects. I do believe that you or a regular contributor meet their conditions: |
Beta Was this translation helpful? Give feedback.
-
|
@raysan5 You could open some kind of fundraiser for Mac and iOS devices. I think the inevitable funds would have been accumulated quickly, given the support of the community. |
Beta Was this translation helpful? Give feedback.
-
|
@Akimpaus sadly I don't see that happening since the patron page shows $88 per month. 🤔 |
Beta Was this translation helpful? Give feedback.
-
|
@Akimpaus I'm afraid that's not easy, as @disketteman commented, Patreon donation are quite low and GitHub Sponsors donations, also public, barely reach the $350 per month. |
Beta Was this translation helpful? Give feedback.
-
|
I think there is a misunderstanding about paying for development on an open-source project. I know this will sound arrogant, but the kinds of funds that are offered simply don't make it worthwhile compared to the actual cost of the tools and extensive effort. Furthermore, the idea is for others to contribute, not add to the workload of the lead developer for whom this is a work of love, not a burden to the needs of others. The missing piece, that Ramon has already identified, is the contribution of another developer or developers who can take responsibility for one or more additional platforms, including future maintenance. It's that simple. And it trumps any armchair quarterbacking on what direction should be taken, what features to support, and what products to integrate with. I will now go quietly and pay attention to what little contributions I am able to make. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for your work on raylib. I think raylib has quietly put itself into a unique position as a defacto gamedev library for many languages that don't have large gamedev ecosystems - even if this was not your intention. While I think OpenGL provides everything an indie dev needs to be successful from technical point of view, my biggest concern is compatibility. Do I think Apple will remove OpenGL tomorrow? No, but I think it would go a long way for raylib to claim support for these 2 large audiences. |
Beta Was this translation helpful? Give feedback.
-
|
doesn't raylib already use ANGLE (if compiled to!) ? This would seem to be the way to go... leverage existing shader based GL code on different graphics backends |
Beta Was this translation helpful? Give feedback.
-
|
@chriscamacho that's correct, ANGLE seems to be supported starting 4.5 based on #2966:
|
Beta Was this translation helpful? Give feedback.
-
|
@disketteman @raysan5 Now there are two possible ways:
|
Beta Was this translation helpful? Give feedback.
-
|
raylib SDL2 backend is already implemented and working |
Beta Was this translation helpful? Give feedback.
-
|
@cos1110 I assume this means that opengl won't be supported by the new SDL3 custom graphics API, not that you can't ask SDL3 for an opengl context and get it. Nothing on that page indicates otherwise. |
Beta Was this translation helpful? Give feedback.
-
|
Side comment: With changes happening on the main trunk now, perhaps it's a good idea to update to a |
Beta Was this translation helpful? Give feedback.
-
@raysan5 I want to challenge the idea of supporting multiple backends, as I don't see the benefits. If the SDL2 support is considered, what's the additional value in supporting and maintaining GLFW and even the rest of the backends? Doesn't SDL2 cover all currently supported platforms plus additional platforms like iOS? And even thought SDL2 provides additional functionality, couldn't link-time optimizer help with that and remove unused functionality? |
Beta Was this translation helpful? Give feedback.
-
SDL has a history of solid backward compatibility layers. https://github.com/libsdl-org/sdl2-compat
Number of open issues != software release cycle.
Vulkan, Metal, OpenGL (ES) and Direct3D are standardized 3D Graphics APIs. GLFW, SDL and FreeGLUT are cross-platform interface for hardware abstraction (using words from the GDC SDL talk). Under normal circumstances you should be able to replace one with another. SDL is the most successful of them. Also SDL supports way more platforms than GLFW. Vulkan can never be a good alternative to GLFW. They are different things.
This recently released video can help. I'm using it for cross platform hobby projects since SDL1.
Unreal doesn't support C# in the first place. |
Beta Was this translation helpful? Give feedback.
-
|
@disketteman My apologies, I meant Unity in the reference to C#. Unreal is all C++ of course. With regard to release cycles the problem is having two release cycles to coordinate, raylib cum SDL, since Ramon does not do bug-fix releases. So a bug in SDL that surfaces as a bug in raylib will take a while to flush out of the system. We have already seen that with miniaudio which has an entirely different cadence and not a formal release process (but very few open issues so changes occur rapidly there). I mentioned Vulkan because it is identified as the successor of OpenGL, though maybe languishing now. My take on all of this is high-level/superficial. I still don't see anything compelling about SDL2/3 in the specific context of raylib though. And my concerns about architectural clashes remain my concerns. Have you considered a kind of pilot demonstration of raylib atop SDL? |
Beta Was this translation helpful? Give feedback.
-
It was Ramon in his post in the first place who considered that, not me! 👇 I did clearly mention that in my original post. I just challenge the idea of multiple backends since SDL might be superior to GLFW and could also replace native implementations.
Please do not be offended, but I have the impression that the conversation is very confusing and we're not getting much from it, so I respectfully decline to continue. |
Beta Was this translation helpful? Give feedback.
-
|
I saw Ramon's statement. I was wondering whether you entertained trying that yourself. I agree. Nothing more to discuss on this. |
Beta Was this translation helpful? Give feedback.
-
I double this. Only one backend. And SDL is the best. But always supporting the official release of the backend, no development versions... raylib needs must not evolve beyond what a linux distro can offer in its packages. This way, it can easily be included in the repositories of all Linux/UNIX distros... |
Beta Was this translation helpful? Give feedback.
-
I've been looking at ways to put my VCrayApp through its paces once I put up the 0.1.0 (beta) release. I will use it in a VCrayVerify project that uses the raylib\examples and demonstrates their compilation and execution using the Visual Studio Development Tools distribution. The recent discussion of |
Beta Was this translation helpful? Give feedback.
-
|
There are problems with glfw in OpenBSD and NetBSD (and with raylib for that matter), but SDL there works nice. Also freeglut works nice but probably less features than the other two? The version of raylib (which is in pkgsrc/wip) that works in NetBSD is 4.0 on top of glfw 3.3.8 but it also has problems with the timing... for the 60 fps to look like 60 fps, you have to set the number much lower... Another thing to look at, the releasing of the mouse button causes a crash (it reports division by zero) in OpenBSD for the C and Pascal application, and in NetBSD only for the Pascal binding ... It seems to me that on these two platforms clang has better support... I do not want to report here bugs, my intention is to say that it might be worth to migrate it to SDL2 ... |
Beta Was this translation helpful? Give feedback.
-
|
I am curious what impact this is intended to have.
PS: I found https://clang.llvm.org/c_status.html and now understand Clang better. It is a different compiler, using LLVM, to pretty much match features of gcc and VC/C++. The support for C Language, by version, is provided at https://clang.llvm.org/c_status.html. The claim to fame has to do with its support by Apple and also the reputation of its diagnostic messages. This is about parsing and front-ends apparently. For example, Visual Studio has Clang compatibility options, so there is a kind of mutual admiration society among the different compiler-development teams. There are differences in optimizations, with the reported comparisons being between gcc and Clang. There is apparently rough parity, each one sometimes being superior among different benchmarks. I don't see any comparison with VC/C++, although that wouldn't figure in efforts directed to *nix platforms. |
Beta Was this translation helpful? Give feedback.
-
|
About the breaking changes or inter-operatibility, it should be possible to just replace GLFW with SDL equivalent in a seamless way for raylib, the main API shouldn't be affected, neither the bindings (that consume the .dll directly) and only the building could require some tweak. That's not the main concern. Main concern is effectively point 3. The redesign and integration cost could be considerable and also maintenance. I'm afraid at this moment I'm not in the situation to assume that cost. Also, SDL3 is on the works, maybe it could be a good moment to consider it again once released. |
Beta Was this translation helpful? Give feedback.
-
|
Preparing for SDL3 would be a great move for raylib. There is a strong
motivation to port SDL on all operating systems defined by a single word:
games. No such things for glfw... and with freeglut is even worse regarding
PR. But it will take a while until SDL3 propagates on all operating
systems...
As for me (that I thought I finally found the final tool for a cross
platform solution and found out that glfw is no such thing). I will try to
adapt raygui to SDL2 as I need it and not willing to wait... This means of
course to start learning SDL2... freeglut is still an option for me...
…On Wed, Jun 14, 2023 at 8:28 PM Ray ***@***.***> wrote:
About the breaking changes or inter-operatibility, it should be possible
to just replace GLFW with SDL equivalent in a seamless way for raylib, the
main API shouldn't be affected, neither the bindings (that consume the .dll
directly) and only the building could require some tweak. That's not the
main concern.
Main concern is effectively point 3. The redesign and integration cost
could be considerable and also maintenance. I'm afraid at this moment I'm
not in the situation to assume that cost. Also, SDL3 is on the works, maybe
it could be a good moment to consider it again once released.
—
Reply to this email directly, view it on GitHub
<#2952 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAGENS3QXBZW73CXC3Y3ISDXLHYEZANCNFSM6AAAAAAVS5EOAM>
.
You are receiving this because you commented.Message ID:
***@***.***>
--
Vasi
|
Beta Was this translation helpful? Give feedback.
-
|
raygui is somewhat prepared for a custom backend port, for example SDL2. Actually it's not implemented yet due to lack of time but I even created an example Please, let me know if you manage to implement it! |
Beta Was this translation helpful? Give feedback.
-
|
Unfortunately, at first glance (well, a few more and a wip mod, learning as I go), freeGLUT is the "dominant species" here, it does not allow polling as raylib needs... it takes control and dictates the terms. It will require one of the two things:
None of it in advantage of raylib :( ... it might work for private solutions that do not need API compatibility and provides only the end product... |
Beta Was this translation helpful? Give feedback.
-
|
The function in case:
But how on Earth is this supposed to work when at the end of the function you close my windows, - and you, free GLUT, say that you do this to allow a new glutInit !? !? !? Practically, you reset everything... |
Beta Was this translation helpful? Give feedback.
-
|
that fgCloseWindows() kills me, everything else is ok until this one... |
Beta Was this translation helpful? Give feedback.
-
|
I'll disable it, to see if I get the functionality I need... |
Beta Was this translation helpful? Give feedback.
-
What do you think about two new functions if(IsKeyDown(KEY_W) cam = MoveCamera(cam, {0.0, +1.0, 0.0});
if(IsKeyDown(KEY_S) cam = MoveCamera(cam, {0.0, -1.0, 0.0});
Vector2 mouse = GetMouseDelta();
cam = TurnCamera(cam, {mouse.x, mouse.y, 0.0}); |
Beta Was this translation helpful? Give feedback.
-
|
// Camera movement
RLAPI void CameraMoveForward(Camera *camera, float distance, bool moveInWorldPlane);
RLAPI void CameraMoveUp(Camera *camera, float distance);
RLAPI void CameraMoveRight(Camera *camera, float distance, bool moveInWorldPlane);
RLAPI void CameraMoveToTarget(Camera *camera, float delta);
// Camera rotation
RLAPI void CameraYaw(Camera *camera, float angle, bool rotateAroundTarget);
RLAPI void CameraPitch(Camera *camera, float angle, bool lockView, bool rotateAroundTarget, bool rotateUp);
RLAPI void CameraRoll(Camera *camera, float angle); |
Beta Was this translation helpful? Give feedback.
-
|
I'd like to see some sort of timing operations to use in games, namely timers that count down (for timed events and triggers) and sequencing of actions (for e.g. cutscenes, animation timing etc.). Example API:
/* (Timers) TimeOut & Interval */
// Timeout executes once
myTimeOut = CreateTimeout(float timeInSeconds, callbackWhenTimeOutCompletes);
StartTimeOut(myTimeOut); // starts the timeout countdown
PauseTimeOut(myTimeOut); // pauses the timeout, useful if your game has a pause screen
ResumeTimeOut(myTimeOut); // resume interval from paused state
FinishTimeOut(myTimeOut); // stops a timeout and executes the callback, also rewinds to be used again with a new StartTimeOut()
StopTimeOut(myTimeOut); // stops/interrupts a timeout without executing the callback, also rewinds to be used again with a new StartTimeOut()
float value = GetTimeOutProgress(); // returns the timeout's current point in time/seconds
DestroyTimeOut(); // clear from memory
// Interval executes over and over looping until cleared
myInterval = CreateInterval(float timeInSeconds, callbackWhenIntervalCompletes);
StartInterval(myInterval); // starts the interval countdown
PauseInterval(myInterval); // pauses the interval, useful if your game has a pause screen
ResumeInterval(myInterval); // resume interval from paused state
FinishInterval(myInterval); // stops an interval and executes the callback, also rewinds to be used again with a new StartInterval()
StopInterval(myInterval); // stops/interrupts an interval without executing the callback, also rewinds to be used again with a new StartInterval()
float value = GetIntervalProgress(); // returns the interval current point in time/seconds
DestroyInterval(myInterval); // clear from memory
/* Note instead of TimeOut and Interval, it could just be "Timer" and have a looping option in the CreateTimer() method to make it loop. TimeOut and Interval just looks more semantic to me. *//* Sequences */
mySequence = CreateSequence(callbackWhenSequenceCompletes, bool looping = false); // looping set to true will cause sequence to loop after executing callbackWhenSequenceCompletes
AddFrameToSequence(mySequence, float timeInSeconds, callbackWhenTimeIsReached); // adds a callback to be executed when sequence reaches that time, all callbacks would have to be added before using StartSequence()
StartSequence(mySequence); // starts the sequence playback
PauseSequence(mySequence); // pauses the sequence, useful if your game has a pause screen
ResumeSequence(mySequence); // resume sequence from paused state
FinishSequence(mySequence); // stops the sequence, also executing any frame callbacks that remain along with callbackWhenSequenceCompletes
StopSequence(mySequence); // stops/interrupts the sequence without executing any more frame callbacks nor callbackWhenSequenceCompletes, also rewinds to be used again with
SetSequenceTimeScale(float value); // a multiplier for the playback speed of the sequence, useful if you want to speed or slow down the playback of the sequence, default is 1.0
float value = GetSequenceProgress(); // returns the sequence current point in time/seconds
DestroySequence(mySequence); // clear from memory |
Beta Was this translation helpful? Give feedback.
-
|
I think those functions would belong to a higher-level engine instead of raylib, they can even be implemented as a separate library in `raylib-extras repo. I think not many raylib users need them. |
Beta Was this translation helpful? Give feedback.
-
|
Software rendering could be cool, but i do not know how difficult that would be to add |
Beta Was this translation helpful? Give feedback.
-
|
Complexity was already analyzed, you can find more info here: #1370 |
Beta Was this translation helpful? Give feedback.
-
|
What about improving Android ecosystem support so we can develop apps more easily? |
Beta Was this translation helpful? Give feedback.
-
|
Some additions to It would be nice to have functions that return In practice, I have avoided these functions altogether and implemented similar functionality in my own project when needed, but it does feel weird that there is this one section the base raylib library that has a caveat like this. Raylib generally keeps thing simple, so I'm not sure if adding a separate set of functions and needing to convey the idea memory ownership is worth sticking in the core library, but I did want to put it out there. I would volunteer to implement these functions you would like to make them a part of 5.0.
The main thing I get when compiling is a bunch of unused function warnings that produce a lot of noise. I can look into getting at least these cleaned up on Linux with GCC/clang. |
Beta Was this translation helpful? Give feedback.
-
Regarding these warnings, would it be a bad idea to add (somewhere around here), something like? |
Beta Was this translation helpful? Give feedback.
-
|
I think this would be better to use something like |
Beta Was this translation helpful? Give feedback.
-
|
Agreed. That's a lot better. |
Beta Was this translation helpful? Give feedback.
-
|
@ashn-dot-dev I agree that suppressing specific warnings that are carefully not worrisome is preferable, especially when using a release-level library. @ubkp How are you using I confess to using option ABOUT OTHER WARNINGS I compile with I address security warnings that arise in conjunction with those Standard C library functions (e.g., As in the above case, I use the recommended secure versions (e.g. These are easier to use without the complexity of |
Beta Was this translation helpful? Give feedback.
-
|
@orcmid Sorry for the late reply, I missed the notification. I'm using I encounter those unused functions warnings only when compiling raylib (the library itself) with I believe the reason for these warnings (aside them not being used) is that those stb functions are declared as |
Beta Was this translation helpful? Give feedback.
-
|
Small question, I recently managed to port the 3 main raylib functions into SDL2 (InitWindow, WindowShouldClose & CloseWindow), would anyone be interested in the code? I’d love to contribute something. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you very much but if the SDL implementation is done, I'll try to make it complete, not partial. |
Beta Was this translation helpful? Give feedback.
-
|
@Clomclem I'm interested, please link them. |
Beta Was this translation helpful? Give feedback.
-
|
I just checked the quaternion to euler functions and they appear to be okay to me? The euler2quaternion function is provably correct but there are a total of 12 solutions for quaternion2euler so i'm not too sure about that one since the formula in there doesn't match the one i know. But it seems to be working fine so i presume its just another solution i dont know |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for the info! Yeah, there was some example that tried to visually illustrate it's functionality converting from Euler to Quaternion and then from Quaternion to Euler but the result were different, so I thought there could be some problem with it.
Yeah, that was it. Good to know it's impossible to achieve! 🤔😄 |
Beta Was this translation helpful? Give feedback.
-
|
Dear Santa @raysan5, I wish Raylib5.0 will be configurable at runtime, including OpenGL version. PS : if you agree with this idea, I will try to contribute. |
Beta Was this translation helpful? Give feedback.
-
|
I'd like to vote against that. IMHO the |
Beta Was this translation helpful? Give feedback.
-
|
@SuperUserNameMan Sorry, that's not planned neither desired, it would imply a huge redesign and I have serious doubts it's even possible to achieve. Also, current |
Beta Was this translation helpful? Give feedback.
-
|
Hello. It seems to me that Raylib is missing a rather simple but important function. Setting a custom cursor image. GLFW has a fairly simple API. |
Beta Was this translation helpful? Give feedback.
-
Well, i'm not sure what @f9x0 had exactly in mind when they wrote "Setting a custom cursor image", but if you want to define a custom "super mario glove" hardware cursor, |
Beta Was this translation helpful? Give feedback.
-
|
@SuperUserNameMan Oh, I see. Well, adding a "custom" system cursor (in addition, or replacement, to the already present) would not be easy to implement at all, also very OS/platform specific and, likely, very out of scope of raylib. |
Beta Was this translation helpful? Give feedback.
-
I don't know about the details regarding how Raylib and GLFW are glued together, but I see that GLFW does have a And if someday Raylib wants to use SDL instead of GLFW, SDL2 also has a As GLFW and SDL function don't work the same way, a wrapping function that would convert a Raylib image into appropriate SDL/GLFW cursor image would be required. I don't know how it would impact the rest of Raylib's API though ... so might be more complicated that what it look. |
Beta Was this translation helpful? Give feedback.
-
|
@SuperUserNameMan Yes, I suggested him to test IMHO, the actual issue that should he should address is the framerate issue. Setting the cursor this way would be just working around that main problem. Edit: editing. |
Beta Was this translation helpful? Give feedback.
-
|
Place a small image in the main loop at the mouse coordinates, before "refreshing" the screen... no penalties... use a global variable that defines which loaded image is used as cursor (mouse pointer) at a moment in time :P (it can be a completely transparent image if the pointer has to be hidden) |
Beta Was this translation helpful? Give feedback.
-
|
My suggestion, I'm not sure if anyone already had it, is to improve raygui to be a basic and fully featured graphical library. Maybe improve/add some aspects like windows, docking, if some button is partially over another, don't click both of them, etc. |
Beta Was this translation helpful? Give feedback.
-
|
Well, there are some problems here that I see. After all, this is a library for games, and the rendering here takes place every frame. For gui, it is better to use rendering as needed. The same problem with event handling. This is a low-level library. To solve the problem of, for example, pressing buttons under each other, you will have to come up with an algorithm yourself. raylib/raygui gives an excellent base, and everything else goes under your "spartan" responsibility. |
Beta Was this translation helpful? Give feedback.
-
|
@luis605 |
Beta Was this translation helpful? Give feedback.
-
|
@Akimpaus raylib provides |
Beta Was this translation helpful? Give feedback.
-
imgui does rendering every frame and it works just fine -- it even has docking -- all the menus and popups in these videos were done with imgui https://www.youtube.com/watch?v=V0fx4hBqbhc https://www.youtube.com/watch?v=Y0DE-nulnls you can go here and scroll down and see some projects people are using it for imgui has implementations for sdl, glfw, opengl and others |
Beta Was this translation helpful? Give feedback.
-
|
@jamesl-github I don't understand the rationale here. If
I look forward to your progress reports on this. |
Beta Was this translation helpful? Give feedback.
-
|
"I don't understand the rationale here." -- DUH -- I put this quote -- "For gui, it is better to use rendering as needed." -- so I was obviously responding to that -- IMGUI shows that for GUIs it does work just fine to render every frame -- raygui does the same and luis605 suggested "to improve raygui to be a basic and fully featured graphical library. Maybe improve/add some aspects like windows, docking, if some button is partially over another, don't click both of them, etc." I'm simply saying if IMGUI can do docking then so can raygui |
Beta Was this translation helpful? Give feedback.
-
|
raygui does not implement docking because it requires an extra level of complexity out of the scope of the project. No plans to implement it. |
Beta Was this translation helpful? Give feedback.
-
|
As far as I have watched in a dozen of game-engine-editors, most of the times developers prefer to implement their own GUI components in an object oriented style, but then leave the technical details to ImGui , to handle everything behind the scenes. For better or worse, this sort of object-oriented design brings clarity into large scale projects. On the other hand I have seen in other large-scale ImGui projects, that are implemented in a linear and monolithic way, and it makes them really hard to understand and to maintain. If you touch something the entire GUI breaks away. That being said, I think that there is a limit where immediate mode GUI is handy. Within small and isolated contexts is safe and easy to use, but as the scale opens up, there are a lot of more difficulties and dangers that lurk. The most plain and easy widget manager, is exactly the same as DatGUI, because is practical and streamlined. It does not do everything, but it does only a few things exactly as it needs to without troubles. You get the window and all widgets are auto aligned one-below-the-other. Another one that is somehow very interesting and works great as a paradigm, is Blender's GUI as you write Python addons. It certainly has restrictions and limitations, but this means that it brings sanity and a streamlined procedure on how the GUI is generated. (I am very sure that Python developers have a great respect for django framework, somehow they got inspired on this). For many years I write Blender GUIs in my addons and not even once I had troubles, while in ImGui I find it very easy to break everything by doing mistakes etc. So the plan for Raylib should be something like this: |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for the feedback, actually, there are no plans for raylib in terms of GUI other than raygui, that I use for my tools and I'm personally very happy and confortable with it. I recommend anyone requiring a GUI library for raylib to try the options out there or build a custom solution. |
Beta Was this translation helpful? Give feedback.
-
|
Sounds like a great list of requirements! Looking forward to seeing what you come up with. |
Beta Was this translation helpful? Give feedback.
-
|
raygui is small, simple, fast, but great is only for people that use it. You can create amazing graphical GUIs with it (even if the tools of @raysan5 looks as flat and boring as possible, having the basic functionality, you can make any kind of control panel you want). Have you ever programmed in "goto style" or "simulated labels"? Of course, combined with procedural programming. Easy to follow and debug. :P Regarding GUI, I develop it alongside application development. I activate a so called "gui designer assistant" where I get a cross hair cursor and its coordinates in the top left of the screen, I "carry" the graphical component which now has the coordinates of the cursor, that meaning that it travels with the mouse, I note down the coordinates and then I fix the component at those coordinates. Very simple, but very helpful. Yes, sometimes the containers can do all, but then I also prefer absolute coordinates, or relative to the position of a rectangle that (falsely) contains them... either way. Funny thing, my first automation application (a PC connected to a washing machine via parallel port) was using Turbo Pascal 7 and BGI unit at 640x480 screen resolution. Without 2D acceleration. I can't remember the name of the cheap graphic card I was using. Many, many years ago. |
Beta Was this translation helpful? Give feedback.
-
|
Hey I am taking a break from professional dev work and spending some time in Mexico and I need a project to work on. I know there has been some talk about converting Raylib to use Vulkan and I am wondering if there has been any progress on this. I have nothing but time on my hands and want start work on something. I know Vulkan support has been on the wishlist, for a while now. What would be the best place to start if someone was going to add Vulkan support? |
Beta Was this translation helpful? Give feedback.
-
|
I did some more research on this and it seems like it would be very hard, basically developing Raylib from scratch |
Beta Was this translation helpful? Give feedback.
-
|
I'd would love to have official iOS support |
Beta Was this translation helpful? Give feedback.
-
|
Now that raylib has been split by platforms, you could implement and test a PLATFORM_IOS yourself, if you wanted. |
Beta Was this translation helpful? Give feedback.
and others
-
Here it is a small wishlist with some possible improvements for future
raylibversions.raylibhas grown A LOT in the last few years and maintenance is getting complicated. For nextraylibrelease I'd like to focus efforts on library maintainability and a better testing of available features. I'd like to also focus on extra libraries and tools onraylibecosystem.Some numbers to take into consideration:
Here it is a list with several elements to review, divided by modules:
module:
rcorercore_desktop.c-PLATFORM_DESKTOP(Windows, Linux, macOS)rcore_android.c-PLATFORM_ANDROIDrcore_native.c-PLATFORM_DRM/PLATFORM_RPIrcore_web.c-PLATFORM_WEBrcore_glfw.c- GLFW library basedrcore_sdl.c- SDL library basedrcore_native.c- Native implementationThere are about 45 functions depending on ~100 GLFW functions, those 45 functions should be reproduced separately for the different platforms/technologies and exposed through the same common API.
SDLlibrary backend:rcore_sdl.c.SDLlibrary is an alternative toGLFW, it's a bit bigger thanGLFWand provides some additional functionality probably not required by raylib but it could be nice to support it as an alternative backend (only the main library, not the auxiliar ones).PLATFORM_RPIbackend. Raspberry Pi Legacy OS (based on Buster) is kind-of deprecated, it uses Broadcom propietary GPU drivers that have been superseeded by an open KMS (PLATFORM_DRM) implementation. RemovingPLATFORM_RPIwill simplify code and maintenance.gladOpenGL extensions library. Use only required expensions per OpenGL version instead of including ALL extensions, it will simplify building times, size and even initialization cost.module:
rlglmodule:
rtexturesImageAPI already provides most of the functionality for 2D software rendeing, including basic shapes, text and even colors math transformations. Old related issue with more details: [core][texture][text][shapes] Software renderer for 2D #1370. Related discussion: Image Drawing Module #2865module:
rl_gputexmodule:
rtextstb_truetypebecause most of the exposed advance API is not used by raylib.module:
rmodelsmodule:
raudiomodule:
raymathcore_quat_conversion.c) but it's not clear enough, it should also be reviewed.module:
rcameraUpdateCameraEx()/UpdateCameraPromodule:
rgesturesREDESIGN: Review the full module. Current implementation should be completely reviewed, it was originally created +7 year ago and it has many issues. Some of the desired features for the re-implementation would be:
TouchEventsinstead ofGestureEvent-> More low-level data processingGestures system has been highly improved recently thanks to the contributor @ubkp.
module:
rayguiRAYGUI_STANDALONE. Current implementation is very coupled to raylib and despite it can be used in standalone mode, implementation is quite difficult. It would be nice to completely decouple it from raylib (in a similar way toDearImGuiorNuklear) and provide separate implementations depending on the required backend (i.e.raygui-raylib.hprovide binding functions for raylib).Fade()function, not depending onguiAlpha, it could be a problem with transparent framebuffers:GuiSliderPro(). Slider is centered onmouse.xposition when mouse pressed over it, it shouldn't move in that case.GuiScrollPanel(). For very big content, the scroll-bar slider could be extremely small, a minimum scroll-bar size should be fixed and scroll movement speed should be probably increased to compensate the minimum size of the slider.TEXT_LINE_SPACING(DEFAULT controls) andTEXT_LINES_SPACING(TextBox control) should be unified in some way.examples
core_3d_camera_free. Displayed controls do not match and camera behaves strangely.models_mesh_instancing. Not working properly on web, issues with the shader but ONLY when uploade to GitHub, it works as expected locally.Isses with several examples on macOS (and maybe other platforms):
custom_frame_control. Produces a very flickery windowmodels_loading. It seems to be buggy at selecting the model, needs some reviewskybox_loading. Doesn't fill the entire window, only the bottom-left quadrantshader_hot_reloading,shader_spotlight. Scaling issue between the mouse and circle positionstop_down_lights. Fully illuminated (no shadows are drawn) until light is movedIssues with several examples not working on
PLATFORM_WEB:models_skybox. WebGL: INVALID_VALUE: disableVertexAttribArray: index out of range _glDisableVertexAttribArray.core_custom_frame_control.core_loading_thread.core_window_flags.raylib_opengl_interop.rlgl_standalone.Issues with several examples not working on
PLATFORM_DRM:models_first_person_maze,core_3d_camera_free). Camera moving uncontrollably.shapes_draw_circle_sector,shapes_draw_ring). Cursor not showing.text_draw_3d,text_input_box). Can't type.audio_module_playing. Crashing the Raspberry Pi after a few seconds.core_custom_frame_control. Flickering and not working.core_vr_simulator. Not rendering.models_skybox. Not rendering.shaders_raymarching. Not rendering.shaders_mesh_instancing. Segmentation fault.rlgl_compute_shader. Not workingraylib_opengl_interop. Not compiling.rlgl_standalone. Not compiling.misc
example.htmlfrom the originalminshell.html. Also, html minification should probably be avoided.There is also a number of open issues and PRs for review.
Those are some possible improvements for next raylib version. It's a long list and it requires lot of work. Note that not all those improvements are expected to be implemented, it's just a reference to follow for future raylib.
As always, contributions are welcome. Feel free to comment.
Beta Was this translation helpful? Give feedback.
All reactions