SDl2 render depends on OpenGL?
on windows/msys2 I have some issues with glew32 link, is normal ? it looks like cmake can't find the lib use glu32 not glw32
sdl2 and sfml samples can't compile with mingw64
Ugh, yeah, I've had a lot of trouble linking glew with CMake as well. What seems to be reliable for me is manually passing the paths to cmake. Eg. when calling CMake pass something like
cmake <...> -DGLEW_INCLUDE_DIR=C:/msys64/mingw64/include/GL -DGLEW_LIBRARIES=C:/msys64/mingw64/lib/libglew32.dll.a
and let libsdl take care of this
what I'm trying to do in to mix lite-xl.github.io wit rmlui and lua ( or wasm )
( + luajit )
@zygon.sisyphus:matrix.org Historically, SDL2 didn't have a way to render arbitrary geometry. However now that this is added just recently, this PR does exactly what you want: mikke89/RmlUi#252
I think libsdl2/sfml can manage lowlevel gx api
and rmlui take care of layout / binding/ logic/...
it will help as crossplatform api
it will be nice to load all as one file ( like asar for electron)
rml/css/lua/resources/...
it can be an squashfs blob inside exe resource and delpoy as one static file
@zygon.sisyphus:matrix.org Appreciate the suggestion. I think abstractions such as those suggested here are better left to clients to implement themselves, or another library itself wrapping RmlUi. We try to be lightweight, and provide quite low-level abstractions that makes it easy to integrate into existing engines and frameworks. This would go against that philosophy.
true, it make sense to have another layer to provide this kind of features.
hi. i'm integrating rmlui into my engine, and i'm seeing that
RenderGeometry() is called sometimes without a CompileGeometry() being called before for that geometry. reading the documentation it seems like this should not happen, am i missing something?
that is, i'm never returning null from
CompileGeometry() yet RenderGeometry() is called
specifically, i'm seeing that when opening the "Element Info" window in the debugger and hovering the mouse on some element, it wants to draw the red rectangle
hmm... looking at the code it seems this is a debugger specific thing, so i guess i can ignore it
@mikke89: i was looking at that because i have a big frame drop while inspecting items in the debugger, and indeed if i make RenderGeometry() do nothing then the fps is much better. but i'm still working on the integration, so maybe it will not be so when i'm finished
i'm doing an homegrown engine using raw Vulkan, not using any pre-made thing
Ah, very cool! I'd love to see more Vulkan integrations, is it mostly for your own sake or do you publish the code somewhere? Perhaps you would be interested in reviewing #236 when it is ready? :)
sure, i can have a look at that, when the fig dissipates a bit ;)
anyway the fps drop seems to be solved now. i was calling map/unmap on the vertex/index buffer every time i had to add a new geometry to them and that was taking a lot of time while now i only map once and keep the pointer around. i knew already that was better and i was planning to do that but i didn't imagine the impact was so big
like from 100 to 1000 fps
@mikke89: hey, i have a question about the scissors: when
EnableScissorRegion(true) is called, will a SetScissorRegion call always come immediately after it, or is it possible for it to set the scissors once, then disable them and then enable again without setting it again, meaning it should use the previous value?
also, is it possible that it calls
SetScissorResion multiple times following a EnableScissorRegion(true)?
@giulio.camuffo:matrix.org We don't really make any guarantees here, but considering these lines are the only place we seem to make a call to the scissor interface, I would say that your assumptions should be reasonable. I don't see any reason for us to change this behavior, but I cannot make guarantees for the future. Maybe add some asserts just in case?
any idea why in windows the samples fonts don't work? i'm getting a freetype error 8 when loading them, and it says they don't contain a unicode or apple roman character map
they work in linux
could it be the different freetype version?
i have 2.10.4 in windows and 2.11.1 in linux
the debugger fonts work
Hey, I'm trying to position my dialog box based on the window edges, using margin only, rather than width and height.
body
{
font-family: LatoLatin;
font-weight: normal;
font-style: normal;
font-size: 2em;
color: white;
background: #f00;
width: 100%;
height: 100%;
}
#box
{
margin-bottom: 80dp;
background: #000;
padding: 1em;
}I got the result in the screenshot above. The black box is not near the bottom at all, it's like it's sticking to the minimal height based on its content.
"bottom" instead of margin-bottom doesn't help. Is there a way to counter this behavior, and is it standard CSS?