Spect 260

[dy]

• clear/trim shortcut
• fixed tail skipping overshoot on identical input/output arrays
• size bumped to 260 (+8b)

[WebReflection]

this looks good to me, but if you find situations that could make some implementation fail, maybe we should add that test too, probably as part of the warm-up/generic checks with assertions?

[dy]

@WebReflection sure, just simple identity transform check:

diff(container, [...childNodes], [...childNodes], before);

It did not fail, but created overshoot.

Another possible test cases I found -

// one in the middle
diff(container, [...childNodes], [...childNodes.slice(0, 2), createNode(0), ...childNodes.slice(3)], before)

// ring
diff(container, [...childNodes], [...childNodes.slice(1), childNodes[0]], before)

// swap tail
diff(container, [...childNodes], [...childNodes.slice(-1), childNodes.slice(1,-1), childNodes[0]], before)

// trim
diff(container, [...childNodes], childNodes.slice(1, -1), before)

I collect them here https://github.com/spectjs/spect/blob/feature/h-meta/test/diff.js

[dy]

Tbh I don't follow the logic of benchmark. Should that be a separate util method for each of the test cases?

[dy]

Ok, just checked libs in spect test suite - snabbdom is broken on prepend #39, all other libs pass all tests except for live input nodes (input array cannot be directly parent.childNodes). If you have an idea how to properly organize tests here - let's do that in a separate PR.