51
General discussions / Re: SFML 3.0.1 released
« Last post by kellanphil on April 29, 2025, 04:49:51 am »Thanks for all the updates SFML version is getting better day by day!
Recent Posts
52
General discussions / Re: SFML releases
« Last post by eXpl0it3r on April 28, 2025, 09:58:18 am »53
General discussions / SFML 3.0.1 released
« Last post by eXpl0it3r on April 28, 2025, 09:57:53 am »We're happy to release a number of bug fixes for SFML 3! 🎉
Changelog
General
- Improved SFML 3 migration guide (#3464, #3478, #3480)
- Improved diagnostics when incorrect library type is found by find_package (#3368)
- Improved diagnostics when C++ language version is too low (#3383)
- Fixed build errors when compiling in C++20 mode (#3394)
- [iOS] Fixed iOS debug build (#3427)
- Removed `-s` suffix for sfml-main (#3431)
- Prevented recreation of UDev target which broke package manager workflows (#3450)
- Fixed bug with installing pkgconfig files (#3451)
- Fixed CMake 4 build error (#3462)
- [macOS] Fixed C++ language version in Xcode template (#3463)
System
Bugfixes
- [Windows] Silenced C4275 warning for `sf::Exception` (#3405)
- Fixed printing Unicode filepaths when error occurs (#3407)
Window
Bugfixes
- Improved `sf::Event::visit` and `sf::WindowBase::handleEvents` (#3399)
- [Windows] Fixed calculating window size with a menu or an extended style (#3448)
- [Windows] Fixed crash when constructing a window from a `sf::WindowHandle` (#3469)
Graphics
Bugfixes
- Fixed `sf::Image` support for Unicode filenames (#3403)
- Ensured `sf::Image` remains unchanged after an unsuccessful load (#3409)
- Fixed opening `sf::Font` from non-ASCII paths (#3422)
- [Android] Fixed crash when loading missing resources (#3476)
Network
Bugfixes
- Fixed comments and address ordering in IpAddress::getLocalAddress (#3428)
- Fixed unsigned overflow in `sf::Packet` size check (#3441)
Contributors
See the full list of contributors on GitHub
Changelog
General
- Improved SFML 3 migration guide (#3464, #3478, #3480)
- Improved diagnostics when incorrect library type is found by find_package (#3368)
- Improved diagnostics when C++ language version is too low (#3383)
- Fixed build errors when compiling in C++20 mode (#3394)
- [iOS] Fixed iOS debug build (#3427)
- Removed `-s` suffix for sfml-main (#3431)
- Prevented recreation of UDev target which broke package manager workflows (#3450)
- Fixed bug with installing pkgconfig files (#3451)
- Fixed CMake 4 build error (#3462)
- [macOS] Fixed C++ language version in Xcode template (#3463)
System
Bugfixes
- [Windows] Silenced C4275 warning for `sf::Exception` (#3405)
- Fixed printing Unicode filepaths when error occurs (#3407)
Window
Bugfixes
- Improved `sf::Event::visit` and `sf::WindowBase::handleEvents` (#3399)
- [Windows] Fixed calculating window size with a menu or an extended style (#3448)
- [Windows] Fixed crash when constructing a window from a `sf::WindowHandle` (#3469)
Graphics
Bugfixes
- Fixed `sf::Image` support for Unicode filenames (#3403)
- Ensured `sf::Image` remains unchanged after an unsuccessful load (#3409)
- Fixed opening `sf::Font` from non-ASCII paths (#3422)
- [Android] Fixed crash when loading missing resources (#3476)
Network
Bugfixes
- Fixed comments and address ordering in IpAddress::getLocalAddress (#3428)
- Fixed unsigned overflow in `sf::Packet` size check (#3441)
Contributors
See the full list of contributors on GitHub
54
Graphics / Re: Text and shaders
« Last post by Me-Myself-And-I on April 28, 2025, 06:04:51 am »I don't think SFML supports this. You might be able to duplicate the effect by drawing a font like this then making something that places each letter along a specified position. You could also make an array ID list to reference the needed image IntRect in relation to the char requested. That is a bit of a hackish method but I think it is basically same to what they did in those example games anyway.
55
Graphics / Text and shaders
« Last post by ThePersonWithAQuestion on April 26, 2025, 01:06:13 pm »Hello there, I'm here again with a new question. To make it short, I want to shade text to like like what you see in this picture:
Every text is different, but what they share is that they have different coloring depending on their pixel position. I wonder what's the best strategy to achieve this. By the way, I'm using ttf fonts.
Every text is different, but what they share is that they have different coloring depending on their pixel position. I wonder what's the best strategy to achieve this. By the way, I'm using ttf fonts.
56
General discussions / Re: Will SFML support HarmonyOS?
« Last post by eXpl0it3r on April 24, 2025, 06:06:56 pm »There are no plans to specifically support HarmonyOS. A lot of the code will like "just work" similar to Android and iOS, but the whole OS API integration surrounding windowing, events, etc. would still need to be implemented.
I don't know anyone of the existing and active maintainer team who would have the necessary time to invest into this, nor do I know anyone with a HarmonyOS device.
There's a good chance that if someone were to spend the time to contribute and test these changes, that it would be accepted.
I don't know anyone of the existing and active maintainer team who would have the necessary time to invest into this, nor do I know anyone with a HarmonyOS device.
There's a good chance that if someone were to spend the time to contribute and test these changes, that it would be accepted.
57
General discussions / Will SFML support HarmonyOS?
« Last post by JasonLeon on April 24, 2025, 05:47:23 pm »HarmonyOS is partly based on Linux, and OpenGL ES 3.2 is available on it.
I'm wondering if the source code of SFML can be compiled on HarmonyOS.
https://developer.huawei.com/consumer/en/doc/harmonyos-references/opengles
I'm wondering if the source code of SFML can be compiled on HarmonyOS.
https://developer.huawei.com/consumer/en/doc/harmonyos-references/opengles
58
Window / Re: Private Joystick Structures
« Last post by eXpl0it3r on April 24, 2025, 02:41:33 pm »SFML's joystick API isn't written to be extended by third-party code, as such the implementation is "private" in the sense.
If you want XInput support, it would be awesome, if you could give the following PR a try and report your testing/findings on the GitHub issue. It's an attempt at adding XInput besides DirectInput: https://github.com/SFML/SFML/pull/3477
If you want XInput support, it would be awesome, if you could give the following PR a try and report your testing/findings on the GitHub issue. It's an attempt at adding XInput besides DirectInput: https://github.com/SFML/SFML/pull/3477
59
SFML wiki / Tutorial - Understanding Text Bounds
« Last post by Hapax on April 23, 2025, 09:31:55 pm »Hello all.
I have written a small tutorial to explain how text bounds work for the wiki:
https://github.com/SFML/SFML/wiki/Tutorial%3A-Understanding-Text-Bounds
Hopefully, this can help to reduce the common confusion to how they work.
Feel free to refer to it when people ask about this topic!
I have written a small tutorial to explain how text bounds work for the wiki:
https://github.com/SFML/SFML/wiki/Tutorial%3A-Understanding-Text-Bounds
Hopefully, this can help to reduce the common confusion to how they work.
Feel free to refer to it when people ask about this topic!
60
General / Using sf::Texture::loadFromFile in a seperate thread.
« Last post by queqomar on April 19, 2025, 11:24:36 pm »Hello everyone,
I'm currently creating a simple project (guess maybe you could call it an engine), for my personal use, that i could reuse in my future applications. So i have thought, that it's a perfect time to start learning some multithreading, and to make the whole project thread-safe.
While working on this, i am using ChatGPT for help.
Basically, Chat has told me, that OpenGL functions that SFML calls internally are not thread-safe, and that loading a sf::Texture object from file is not thread safe also, and loading it, in a separate thread, might cause undefined behavior - because of calls to the OpenGL functions and context switching.
In some of my previous projects i had already used separate thread for loading textures from file, while having a loading screen. I have never had any problems with it, and the loading has succeeded every time (but i have never tested it outside my PC).
The Chat, has also proposed to me a solution to this - he said, that i should load a sf::Image object from file in a separate thread, but then, in a main thread use loadTextureFromImage - in his opinion it would fix this problem.
I'm not sure i can always trust chat with this - especially since i have never had such a problem.
So my question is, is A.I. right about that?
Should i not use a separate thread that loads textures, is it unsafe?
If so, then what are other limitations of multithreading in SFML that i should be aware of?
Is Chat's potential fix to the problem good?
While browsing SFML's texture code on the GitHub, i have found a comment, that says:
"Force an OpenGL flush, so that the texture will appear updated in all context immediately (solves problems in multi-threaded apps)"
I have never encountered an issue with this approach, but I'm concerned that it might not work universally, and since i have not encountered any reproducible errors i don't have any example code to share.
Thanks in advance for you help and advice!
I'm currently creating a simple project (guess maybe you could call it an engine), for my personal use, that i could reuse in my future applications. So i have thought, that it's a perfect time to start learning some multithreading, and to make the whole project thread-safe.
While working on this, i am using ChatGPT for help.
Basically, Chat has told me, that OpenGL functions that SFML calls internally are not thread-safe, and that loading a sf::Texture object from file is not thread safe also, and loading it, in a separate thread, might cause undefined behavior - because of calls to the OpenGL functions and context switching.
In some of my previous projects i had already used separate thread for loading textures from file, while having a loading screen. I have never had any problems with it, and the loading has succeeded every time (but i have never tested it outside my PC).
The Chat, has also proposed to me a solution to this - he said, that i should load a sf::Image object from file in a separate thread, but then, in a main thread use loadTextureFromImage - in his opinion it would fix this problem.
I'm not sure i can always trust chat with this - especially since i have never had such a problem.
So my question is, is A.I. right about that?
Should i not use a separate thread that loads textures, is it unsafe?
If so, then what are other limitations of multithreading in SFML that i should be aware of?
Is Chat's potential fix to the problem good?
While browsing SFML's texture code on the GitHub, i have found a comment, that says:
"Force an OpenGL flush, so that the texture will appear updated in all context immediately (solves problems in multi-threaded apps)"
I have never encountered an issue with this approach, but I'm concerned that it might not work universally, and since i have not encountered any reproducible errors i don't have any example code to share.
Thanks in advance for you help and advice!