Test for A8-on-R8 swizzle being applied to the wrong tex unit.

[kdashg]

Firefox bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1503813

[kenrussell]

Sorry for the delay reviewing. LGTM

[lexaknyazev]

Is it expected that Chrome 72 doesn't pass this test? Namely:

Tests that glActiveTexture and glBindTexture work as expectedSpecifically texture targets are per active texture unit.

On success, you will see a series of "PASS" messages, followed by "TEST COMPLETE".

PASS getError was expected value: NO_ERROR : 
PASS getError was expected value: NO_ERROR : 
PASS getError was expected value: NO_ERROR : 
PASS getError was expected value: NO_ERROR : 
PASS getError was expected value: NO_ERROR : 
PASS should be 0,192,128,255
PASS should be 128,64,255,255
PASS should be 192,255,64,255
PASS should be 200,0,255,255

Swizzle applied to correct ActiveTexture index
PASS getError was expected value: NO_ERROR : 
FAIL window.control should be 10,20,30,40. (was 0,192,128,255)
FAIL window.after_swizzle_applied should be 10,20,30,40. (was 0,192,128,255)
PASS successfullyParsed is true

TEST COMPLETE

It seems that drawArrays here has no effect on the subsequent readPixels call.

WebGL/sdk/tests/conformance/textures/misc/texture-active-bind.html

Lines 168 to 174 in c2419a2

	window.control = new Uint8Array(4);
	window.after_swizzle_applied = new Uint8Array(4);
	gl.bindTexture(gl.TEXTURE_2D, tex0_rgba8);
	gl.drawArrays(gl.POINTS, 0, 1);
	gl.readPixels(0, 0, 1, 1, gl.RGBA, gl.UNSIGNED_BYTE, window.control);
	shouldBeString("window.control", tex0_rgba8_data.toString());

[jdarpinian]

We have a bug filed for this already: http://crbug.com/931006

[lexaknyazev]

@jdarpinian Thanks for the link!
Further investigation shows that Chrome doesn't render points unless vertex shader sets gl_PointSize.
So, this test should be fixed here (#2819), and the spec should probably be more strict (#2818).