Fix VCR to record simultaneous keys+mouse #1825
Conversation
Catch up with upstream.
| var key:Int; | ||
| var x:Int; | ||
| var y:Int; | ||
| var possibleKeys = [8, 9, 13, 16, 17, 18, 20, 27, 32, 33, 34, 35, 36, 37, 38, 39, 40, 45, 46, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 109, 110, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 186, 187, 188, 189, 190, 191, 219, 221, 222, 301]; |
There was a problem hiding this comment.
This is a bit silly... You could probably use FlxKey.toStringMap.keys() (as long as -1 and -2 are excluded).
There was a problem hiding this comment.
Also backslash and graveaccent, so you don't bring up the debugger and throw off other tests? This test ended up throwing off the flxbutton test because it moved the mouse, so I guess they're not set up to be independent, exactly.
There was a problem hiding this comment.
Really? That's weird.. There should be a FlxG.resetGame() call after each test class at least.
There was a problem hiding this comment.
And I went to do it as you said, but that returns an iterator. I then learned a new trick, thanks! :)
There was a problem hiding this comment.
There was a problem hiding this comment.
An iterator can be converted to an array fairly easily.
There was a problem hiding this comment.
I know that now ("I then learned..."), see my latest commit. :)
| @@ -14,6 +17,7 @@ class FlxButtonTest extends FlxTest | |||
| { | |||
| button = new FlxButton(); | |||
| destroyable = button; | |||
| FlxG.mouse.playback(new MouseRecord(0, 0, FlxInputState.RELEASED, 0)); // put the mouse in the upper-left corner to prepare for highlight testing | |||
There was a problem hiding this comment.
I would not have thought it so, except that after starting from the current dev branch, this test was failing. After resetting the cursor, it was succeeding.
There was a problem hiding this comment.
I guess resetGame() either doesn't run between the tests, or it does, but doesn't touch the cursor...?
There was a problem hiding this comment.
Hm, yeah.. I thought resetGame would completely recreate all instances in FlxG, guess it's actually far from that. All this global stuff is pretty annoying for unit testing... :/
We should at least handle the mouse reset in FlxTest#resetGame(), so it's not just a random hack, but happens consistently. Perhaps we can even manually call FlxG.init(), so things really get reset (new FlxMouse instance for example).
There was a problem hiding this comment.
That's interesting. FlxTest#resetGame() calls FlxG.resetGame(), which eventually should mean hitting FlxG.reset(), which calls input.reset(), which ought to, ya know, reset the inputs, I guess. But actually FlxMouse's reset() just resets the mouse button states. I guess that makes sense, otherwise you might have a weird cursor jump for the user, but I'm not sure how resetGame() gets used in the wild anyway.
There was a problem hiding this comment.
Could you be a bit more specific than "causing trouble"?
There was a problem hiding this comment.
Sorry, see edit. It was the trouble I'd already described (crashing, other tests failing) but there's more detail now. Also re: hang vs crash, sometimes there's a window that comes up, with a Windows crash, or the first time I saw it, maybe I just didn't see the window, but from the CLI, the test runner was just sitting there forever (same output minus the Error:...)
There was a problem hiding this comment.
I just tried a new approach:
Lib.current.stage.removeChild(FlxG.game);
FlxG.game = null;
Lib.current.stage.addChild(new FlxGame(640, 480, FlxState, 1, 60, 60, true));Which works better: the FlxButtonTest and the others that broke now pass. However, it still makes the FlxReplayTest fail with accumulated keys. I find this strange, but we're getting closer...
I guess there's no way to clean up FlxG's static stuff, or clean up at all, since it's meant to just evaporate when the game quits, right? As TestMain.hx says,
// Flixel was not designed for unit testing so we can only have one instance for now.
So we have to figure out how to clean up that one instance...
There was a problem hiding this comment.
That seems to have a pretty big impact on the runtime of the tests. Perhaps it's best to keep it as simple as we need it for now - we could just create a new FlxMouse instance manually. There's even already a setter that handles all the details of that since we needed that in flixel-ui at some point.
There was a problem hiding this comment.
Or, I could just make the final playback frame in the flxrecord test put things back where they were. It's easier and amounts to the same thing in the end, resetting it to 0,0 and with no buttons down.
… than at the start of the button test.
|
Tests pass locally. |
|
Thanks for your patience! |
Fix VCR to record simultaneous keys+mouse, closes HaxeFlixel#1739 (HaxeFlixel#1825)
C.f. #1740, this is a fresh branch redo. Tests pass locally on cpp.