This makes them consistent with window and framebuffer size queries,
that are also not cached but queried every time. It fixes a case where a
global UI scaling change in the OS triggered a viewport event but the
event didn't actually have the DPI scaling value updated.
It doesn't however handle actual explicit DPI change events yet, that's
another nightmare altogether.
There is one already, and I'm not going to use because it's bloated.
Plus there's EmscriptenApplication that doesn't have to pay for the
extra overhead of wrapping HTML5 APIs in SDL APIs.
Same as the corresponding change in Corrade, this allows each function
to explicitly specify its dependencies, making it no longer depending on
what a particular Emscripten version decides to include by default, or
forcing users to painstakingly fill the EXPORTED_FUNCTIONS array when
linking the final executable.
It also allows the code to eventually get conditionally included or not
with preprocessor branches, for example to not include environment
queries for code that won't ever access Node.js console.
This makes the minimal supported Emscripten version 1.39.5. With some
more effort this could be changed to 1.38.27, but I don't think anybody
needs that.
AsciiToString is not included by default on 3.1.21+ and including it is
basically impossible on the library side because I don't think they
fixed the case of supplying multiple DEFAULT_LIBRARY_FUNCS_TO_INCLUDE
options on the command line in order to concatenate those lists yet.
Also, given that UTF8ToString is probably already used in other places
since it's included by default, using AsciiToString would only mean
inflating the JS code.
I was abusing the API and passing a negative pitch there to not have to
invert the image by hand. It stopped working in 2.23 when they hardened
the argument checking and, while working correctly, this feature was
accidental and undocumented.
Unfortunately it broke silently, because the API returned nullptr and
SDL_SetWindowIcon(..., nullptr) is then resetting an icon to nothing. So
I'm adding an internal assertion there now. Hopefully it doesn't start
blowing up for some reason again, heh.
The code right below is querying the `log` option, which wasn't added.
Becomes a problem when Magnum is compiled w/o GL support, e.g. for a
custom WebGPU renderer.
I'm getting kinda pissed at the strange defaults. Should I be using a
different toolkit altogether because SDL IS FOR GAMES ONLY? Given how
many bugreports and complaints there is about "Dosbox blocking
screensaver" I'm beginning to think it's SDL's fault, not mine.
Let the users control what they want, ffs, don't enable problematic
features like blocking powersave or disabling compositor by default.
Those should be a runtime option anyway, similarly to how video players
block powersaving only when *an actual video is playing*.
A large portion of the needed changes was in the previous commit
already, this does just the remaining part, in particular ensuring EGL
is linked and SDL is told to use EGL as well -- GLFW was told so in the
previous cleanup commit already.
These two options were mutually exclusive, and both were doing the same
thing -- switching to EGL on desktop GL, or switching away from EGL on
GLES. That made all logic vastly more complicated than it should be, and
unfortunately it took me half a decade to realize that. The new logic is
significantly simpler everywhere.
As usual, the old options are still recognized by CMake on a deprecated
build (with a warning), and are still exposed both as CMake variables
and a preprocessor define. But the logic for them was quite complicated,
so I don't guarantee all cases are covered.
I also tried to clean up the dependent CMake options to allow building
GLX and WGL apps on GLES independently of whether EGL is used, but it's
quite a mess due to the limitations of CMake < 3.22. Build directories
that have the options switched randomly over a long time might start
misbehaving, but the initial build should work well.
For quite a while, setSwapInterval() was reporting that "swap interval
was ignored by the driver". Since I used to have that behavior ages ago
on a NVidia Optimus machine (where it was just *impossible* to have
VSync, imagine that!!), I assumed it was a similar wart in Mesa and
didn't bother looking into it.
It turns out, however, that calling setSwapInterval(1) may result in
SDL_GL_GetSwapInterval() returning -1 instead of 1, thus helpfully
enabling late-swap behavior for me. Since -1 != -1, the code treated
that the same as if SDL_GL_GetSwapInterval() returned 0 (which was the
case with NV Optimus having broken VSync), but it's not an error in
fact.
Which allows to get rid of a now-unneeded ArrayView include in the
header. Also, now that we're returning a StringView, it's useful to
have the view always null-terminated. In SDL2 and Emscripten it was
already like that, GLFW needed a minor change.
Except CGL, iOS, AndroidApplication and AbstractXApplication which are
either too crappy or don't have the needed scaffolding for specifying
context flags yet.
We no longer have to use sizeof("...") to avoid useless strlen calls or
check for nulls because the StringView APIs are ACTUALLY SANE and not a
full of nasty surprises and performance / security pitfalls like with
both the C and C++ standard library functions. Good riddance.
This removes one unnecessary allocation from each application startup.
In some cases of the windowless apps the Platform::GLContext could be
put directly into the class, in other cases it had to be wrapped in an
Optional because we need delayed construction and/or earlier
destruction.
So in case it touches the GL state in some way, it doesn't do that on an
already destroyed context. The windowless apps do this all implicitly
due to the WindowlessGLContext encapsulation.
Right now only the command-line variant of it was checked. Since on
some platforms this requires the app to explicitly request a debug
context, the app needs to handle the case when it's passed via a
Configuration as well.
Disabling engine startup log or modifying enabled extensions /
workarounds from the application side was one of the common pain
points and this should *finally* solve the problem. This Configuration
is now inherited by the usual Platform::*Application::GLConfiguration /
Platform::Windowless*Application::Configuration classes people are used
to, so for the end user it's just as if these classes got a bunch new
options.
Having this, I also extended the ContextGLTest to verify that the
Configuration and command-line options do what's expected because that
hadn't automated tests until now. The test is mostly a copy of what I
did for Vulkan already, nothing special. Additionally all
Platform*ApplicationTest executables gained a new --quiet option to
verify that the GL::Context::Configuration subset gets correctly passed
from the Application code, because that's something we can't really
verify in an automated way.
These are in most cases the only strings that are used, and I don't
think having to call std::strlen() for each of them is a good idea if
we don't need to.
There's four more new cursors and the _cursors array was too small.
At first I got confused because I thought the assertion on top is done
against the CursorMap, which didn't contain the Hidden cursors. So to
avoid confusing myself again in the future, I moved the assert after the
special cases and made both arrays the same size since it doesn't make
sense to have always-empty fields in there.
Similar change is done in Sdl2Application, and an assert is added to
avoid a nondescript crash if the window is not created yet.
It was printing 0 before, which isn't correct. Also why not print both
values? Printing just the first one would hide issues where the second
is accidentally 0 or some other wrong value.
Instead of first entering the main loop, processing events etc. This
also makes it finally possible to exit the application cleanly, with all
non-global destructors executed as well.
If the application is constructed with delayed window creation, the
warning would be printed on each context creation attempt / call to
dpiScaling(const Configuration&), which is silly.
Vladimír Vondruš