Last call for 2.0.6 bugs!


#1

We are working to finalize a 2.0.6 release!

We don’t promise to fix everything, but if there’s a bug that’s been bothering you and you feel hasn’t gotten the attention it deserves, we need to hear about it now so we can triage appropriately.

Please reply to this topic with any must-fix issues you have!

(Links to a specific report at https://bugzilla.libsdl.org/ are optional, but greatly appreciated.)

Thanks!


#2

#3

Hi,

Please review the patch from https://bugzilla.libsdl.org/show_bug.cgi?id=3637.

It fixes an annoying bug in ScummVM on X11 where pressing Alt+Enter switches to fullscreen and immediatly back to windowed mode due to an extraneous SDL_KEYDOWN event (https://bugs.scummvm.org/ticket/9617) easily reproducible using the code from ScummVM PR#937.


#4

Hi, I’ve got a bug I’ve been wrestling with for months now: Haptic feedback no longer works in OS X release builds

https://bugzilla.libsdl.org/show_bug.cgi?id=3659


Haptic feedback no longer works when built with modern versions of Xcode. Haptic feedback works normally in debug builds, but release builds without a debugger attached no longer work.

Xcode 7.2.1 and previous versions work correctly.
Xcode 7.3 and onward all exhibit this issue.


#5

Would you please change the value of SDL_WINDOW_VULKAN to 0x10000000 for compatibility with Tizen’s sdl fork?
See https://review.tizen.org/git/?p=platform/upstream/SDL.git;a=blob;f=include/SDL_video.h;h=4c60c2703b02b921afdef5a804729f47de06713e;hb=HEAD#l113
and https://bugzilla.libsdl.org/show_bug.cgi?id=3591#c37


#6

This has been waiting for a couple of releases.
Overscan mode/hint for Logical Renderer
https://bugzilla.libsdl.org/show_bug.cgi?id=2799

Thanks,
Eric


#7

Would you please change the value of SDL_WINDOW_VULKAN to 0x10000000 for compatibility with Tizen’s sdl fork?

I had mentioned this before, but we aren’t willing to promise ABI compatibility with this patch until we’ve reviewed it.

That being said, Vulkan support (presumably from this patch) is definitely going into 2.0.6 in some form.

But it feels like they chose such a large value so it wouldn’t conflict with SDL development; has this shipped to the public in any meaningful way yet? If not, honestly, I’d rather they change it in Tizen to 0x100000.


#8

I’m not very familiar with the tizen release process, however, from what I can tell, it has shipped. The version that appears to be the latest release/prerelease is Tizen 3.0 M2 (see https://source.tizen.org/release). I downloaded the source RPM for SDL from http://download.tizen.org/releases/milestone/tizen/3.0.m2/3.0.m2-common/latest/repos/x86_64-wayland/source/SDL2-2.0.4-2.2.src.rpm. In the included diff, SDL_WINDOW_VULKAN is set to 0x10000000 (on line 39436).

Also, in the release notes for 3.0 M2, they mention that their version of SDL supports Vulkan.


#9

https://bugzilla.libsdl.org/show_bug.cgi?id=3568#c3

Hey, please fix this one and let us, the Visual studio pleb, using some on the new 2.0.5 function without editing (badly) the .lib ourself. Can’t link when using SDL_CreateRGBSurfaceWithFormatFrom() :frowning:

Thank’s a lot !


#10

Hello,
This is not critical issue and not only SDL 2.0.6 related but any SDL versions.

Unknown behaviour in src/events/SDL_mouse.c:66:26: runtime error: shift exponent -1 is negative
See more into in https://bugzilla.libsdl.org/show_bug.cgi?id=3593


#11

https://bugzilla.libsdl.org/show_bug.cgi?id=3568#c33

I think this should get resolved when Sam builds the 2.0.6 libraries…? I don’t think there’s any changes we need to make, it was just a building error last time…I think.


#12

We’ll see, I looked in the source and it “seem” good (not an expert at all), so probably build error or lib older than the dll. Just wanted to rememb it because it annoy me and can really annoy other sdl newbies. Thx


#13

Hi,

I was wondering if SDL would ever be able to use floats in the renderer. It could fix a lot of animations smoothness problems I think.


#14

I was wondering if SDL would ever be able to use floats in the renderer.

This was a design mistake in 2.0; we were thinking about mapping directly to pixels, but you can see places this causes problems with rotation, etc.

I don’t think we’re going to change this for 2.0.6, but we should either correct this in 2.1 or introduce replacement APIs for 2.0.

(There might be a patch somewhere on Bugzilla for this, too, if I recall.)


#15

It will be a great improvement, thanks!


#16

Related to this, mouse event coordinates (scroll, motion) should be float as well, so they can express exact pixels on a 200% scaled highdpi display (for example).

Looks like there is a patch for float scrolling: https://bugzilla.libsdl.org/show_bug.cgi?id=2293

Maybe this is more post 2.0.6, but thought I’d mention it anyway.


#17

Could antialiasing be added as a render hint? Currently i can’t find any pure SDL2 antialiasing librairies or functions that could make the job. I guess it would’nt be added in 2.0.6 but I’m asking anyway


#18

I’d love if this could be resolved:
https://bugzilla.libsdl.org/show_bug.cgi?id=3640

It is not, by all means, critical, but also shouldn’t take too long to fix :slight_smile:


#19

I’d really like to get android cmake fixed.

I collected/updated/made some fixes to get a working SDL version for my self that fixes these issues:
https://bugzilla.libsdl.org/show_bug.cgi?id=3509
https://bugzilla.libsdl.org/show_bug.cgi?id=3267
https://bugzilla.libsdl.org/show_bug.cgi?id=3195

See https://github.com/suikki/simpleSDL for more details, a sample project and a patch. Or should I file a separate bugzilla issue?


#20

I collected/updated/made some fixes to get a working SDL version for my self that fixes these issues:

I’ve applied your patch and closed out the first two bugs, but I suppose https://bugzilla.libsdl.org/show_bug.cgi?id=3195 is still an open issue, since our example project is still using Ant.

–ryan.