Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • 10:18
    magenbluten starred mosra/magnum
  • 10:13
    magenbluten starred mosra/corrade
  • 08:25
    williamjcm commented #74
  • Dec 13 21:39
    codecov-io commented #406
  • Dec 13 21:39
    Build #2371 passed
  • Dec 13 20:44
    mosra commented #74
  • Dec 13 19:13
    roig opened #406
  • Dec 13 17:44
    williamjcm commented #74
  • Dec 13 04:24
    LeiFeng1000 starred mosra/magnum
  • Dec 13 01:35
    zhaoxiaohan starred mosra/magnum
  • Dec 12 22:16
    mosra closed #71
  • Dec 12 22:16
    mosra commented #71
  • Dec 12 22:11
    mosra labeled #71
  • Dec 12 22:08
    mosra commented #73
  • Dec 12 22:07
    mosra closed #72
  • Dec 12 22:07
    mosra commented #72
  • Dec 12 22:05
    mosra labeled #72
  • Dec 12 21:58

    mosra on ports-next

    fluidsimulation3d: Fix color op… fluidsimulation3d: fix some war… fluidsimulation3d: change heade… and 36 more (compare)

  • Dec 12 21:55

    mosra on master

    doc: mark new examples as such. fluidsimulation2d: stupid me me… fluidsimulation{2d,3d}: minor c… and 2 more (compare)

Nghia Truong
@ttnghia
@mosra I realize that all your webgl examples cannot detect the correct pointer position?
for ex., in the image above, the mouse pointer is far away from its detected position (green circle)
Pablo Escobar
@pezcode
works ok for me
Nghia Truong
@ttnghia
I tried other example such as the bullet one (https://magnum.graphics/showcase/bullet/) and have the same problem
Pablo Escobar
@pezcode
are you on a hidpi screen?
Nghia Truong
@ttnghia
I'm not sure, but I guess hidpi could be the reason
because of this:
image.png
image.png
image.png
when the mouse pointer is near the upper-left corner, the error is smaller
the farther from upper-left, the larger mouse position error
Marco Melorio
@Melix19
Same problem here (macOS 2x dpi)
Nghia Truong
@ttnghia
I found this, but I don't know how to test https://www.khronos.org/webgl/wiki/HandlingHighDPI, waiting for mosra...
Guillaume Jacquemin
@williamjcm

growable arrays are now in corrade master

Noice.

Alan Jefferson
@alanjfs
Regarding cursor in WebGL, works ok for me with 1.5x dpi, on Windows
francesco-cattoglio
@francesco-cattoglio

they did, but then Spectre happened

Is that really the reason for which the support was disabled?

That doesn't make much sense to me, I don't remember multithreading having a big impact on being able to exploit the branch predictor
Vladimír Vondruš
@mosra
what the heck ... just yesterday i merged a hidpi-related fix for some other example, but i'm pretty sure these worked correctly on my hidpi screen
well imgui worked, at least, didn't try the cursor (does imgui work for you?)
bullet might be broken too actually, since i don't remember fixing it :D
@francesco-cattoglio shared memory got disabled, and without that multithreading is pretty useless
francesco-cattoglio
@francesco-cattoglio
Oh, I see!
Well, I hope we get multithreaded webasm back soon-ish
Since I might get back to "interesting" software development soon enough! ^^
Marco Melorio
@Melix19

does imgui work for you?

Yep

Nghia Truong
@ttnghia
imgui/box2d/picking work correctly for me
wait
imgui in this example(https://tmp.magnum.graphics/fluidsimulation2d/) also works correctly
so, somehow, the mouse coordinate passed into mouseEvent is not correct
Vladimír Vondruš
@mosra
nono, it's just the coordinate scaling for the cursor ... on the web the window size is different from framebuffer size and that's what breaks it
i'll fix it :)
Nghia Truong
@ttnghia
:smile_cat:
Guillaume Jacquemin
@williamjcm
I tried that example, and the lowest framerate I got was 12.3 FPS.
Vladimír Vondruš
@mosra
not bad! i got 6
Nghia Truong
@ttnghia

I tried that example, and the lowest framerate I got was 12.3 FPS.

I got similar

Guillaume Jacquemin
@williamjcm
I thought my Ryzen 3700X would be better than that, but if the example is single-threaded, then I understand.
The 3700X has 8 cores and 16 threads, and it's a beast in apps that can do proper multithreading (mostly workstation loads).
Jonathan Hale
@Squareys

https://blog.squareys.de/cmake-fixup-bundle-cpack/

In case anybody ever comes across the same issue

Guillaume Jacquemin
@williamjcm
I'll keep it in mind. Sounds useful.
Jamil Halabi
@jay3d
#if !defined( __clang__ ) && !defined( __GNUC__ )
        // Hey Microshaft!! Are you watching this? This is your buggy-ass compiler!!
        ( &MatCapShader )->SetGhostingOpacity( GhostingOpacity );
#else
        MatCapShader.SetGhostingOpacity( GhostingOpacity );
#endif
Guillaume Jacquemin
@williamjcm
Wait, what ? o_O
Does it really refuse to let you access the object without going through a pointer dereference ?
Jamil Halabi
@jay3d
No it just says syntax error '.'
Guillaume Jacquemin
@williamjcm
Weird.
Jamil Halabi
@jay3d
I think MSVC is its past life was a GWBASIC compiler :D
I'm using Clang-cl because MSVC OpenMP is only at 2.0 which is ridiculous!
The only free MSVC compatible compiler that supports OpenMP 4.5+ is Clang-cl
Jamil Halabi
@jay3d
custom reduction is not supported in OpenMP 2.0
#ifdef __clang__
#warning "Using Clang compiler"
#pragma omp declare reduction( merge:Uint32Array : omp_out.insert( omp_out.end(), omp_in.begin(), omp_in.end() ) )
#endif

        Uint32Array changedVertices;
        changedVertices.reserve( vertexCount );

#ifdef __clang__
#pragma omp parallel for reduction( merge : changedVertices )
#else
#pragma omp parallel for
#endif
        for ( auto i = 0; i < vertexCount; ++i )
        {