MacOS: fix render issue in obs-32 branch

[sandboxcoder]

Description

• Fix issue in Obs-32 branch where create_iosurface uses kIOSurfaceIsGlobal: YES but then immediately called CFRelease causing IOSurfaceLookup in both write_iosurface and node-window-rendering to return NULL.
• Fix pre-existing issue (while we are in here) where we'd have OpenGL glClear errors being sent to the logs. Add checks for 0x0 sized buffer.
• Add optional DEBUG_WRITE_IOSURFACE code path which will dump the IOSurface contents into a .png file which can be helpful determining if backend is properly rendering (because if we have pixels, then problem is upstream in NWR/Desktop).

Motivation and Context

Depends on this PR: #734. Fix rendering issue for macOS in obs-32 branch

How Has This Been Tested?

Verified I see images rendered during dual output etc on Streamlabs Desktop in obs_merge_32.1.1 locally

Types of changes

• Bug fix (non-breaking change which fixes an issue)

Checklist:

• My code has been run through clang-format.
• I have read the contributing document.
• My code is not on the streamlabs branch.
• The code has been tested.
• All commit messages are properly formatted and commits squashed where appropriate.
• I have included updates to all appropriate documentation.

[Copilot]

Pull request overview

This PR targets macOS OpenGL rendering by keeping the created IOSurfaceRef alive (instead of releasing it immediately) so IOSurfaceLookup-based consumers don’t see NULL, and by tightening present/readback behavior to avoid benign GL errors.

Changes:

• Persist and manage the lifetime of the created IOSurfaceRef (store on swapchain/windowinfo; release on teardown).
• Adjust swapchain/present readiness logic and add optional debug IOSurface dumping support.
• Remove UniformTypeIdentifiers dependency in the IOSurface debug helper (use "public.png" UTI string).

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

File	Description
libobs-opengl/IOSurfaceHelper.m	Removes UTType dependency and updates PNG type usage for debug image export.
libobs-opengl/gl-cocoa.m	Stores/releases IOSurfaceRef, updates swapchain load/present readiness, and changes IOSurface readback path.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 3 comments.

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated no new comments.

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

Comments suppressed due to low confidence (1)

libobs-opengl/gl-cocoa.m:723

• If IOSurfaceCreate fails, this function still returns swap->wi->surfaceID, which may be a stale, non-zero ID from a previous successful call. That makes failure indistinguishable from success for callers. Return 0 on failure and only update swap->wi->surfaceID when a new surface was actually created.

    IOSurfaceRef surfaceRef = IOSurfaceCreate((CFDictionaryRef) surfaceAttributes);
    if (surfaceRef) {
        swap->wi->surfaceID = IOSurfaceGetID(surfaceRef);
        if (swap->wi->surfaceRef) {
            CFRelease(swap->wi->surfaceRef);
            swap->wi->surfaceRef = NULL;
        }
        swap->wi->surfaceRef = surfaceRef;
    } else {
        blog(LOG_ERROR, "create_iosurface IOSurfaceCreate failed");
    }

    [surfaceAttributes release];

    return swap->wi->surfaceID;
}

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

[aleksandr-voitenko]

When IOSurfaceCreate fails, this clears surfaceID but leaves the previous surfaceRef alive. Since write_iosurface now gates on surfaceRef, a failed create or 0x0 resize can still keep writing into the stale IOSurface even though callers see ID 0. Please release/null surfaceRef here too, and ideally explicitly return early for width == 0 || height == 0.

[sandboxcoder]

I thought bout that but received this defect from copilot which made me think it's better to keep the existing one? Updated in 6c47cae, early reject if width/height is 0