Ryan C. Gordon

New current plan: I'm going to bump this to 3.6.0. Getting down to the wire, don't want to make big changes and additions at this point.

I need to dig into the implementation to see if there are obvious wins we can pick up. I suspect if nothing else, it could be worth dirtying up some...

Hmm, maybe we add an API: ```c const SDL_PropertiesKeyValuePair keys_and_vals[] = { // last field would probably be a void*, to hold pointers, numbers, and bools, and likely would need...

I took a run at my idea; if this turns out to be a terrible idea, no foul at just closing the PR unmerged, though.

> Would a SDL_MakePropertiesReadOnly(SDLPropertiesID) call also work? It could save on locking once the data is ready, but there's still a giant flurry of malloc and LockMutex calls to build...

One last note before I move on to other 3.4.0 stuff: I just realized that SDL_IOFromFile isn't a thing that funnels into a generic SDL_CreateIOStreamWithProperties() function. There the problem is...

Right now we grow as necessary (preallocating 4 buckets), which is probably the right thing to do for SDL_Properties, since they tend to only have a handful of items anyhow...but...

So my thinking, which I haven't been able to test yet, is that moving to a new screen either isn't calling `-[NSView resetCursorRects]`, or our implementation... ```objc - (void)resetCursorRects {...

Tried this on a MacBook Pro with a 4K display plugged into its HDMI port, and I couldn't reproduce the issue with testcustomcursor.c. Is there a build of your project...

> is that moving to a new screen either isn't calling -[NSView resetCursorRects] (This definitely wasn't true, btw. Cocoa is calling this message _way_ more than I expected it to.)