[css-grid] Doubts on worthiness of grid-auto-flow:stack

I just slightly reworked the auto-placement algorithm to define the
behaviors of the three placement algorithms separately, and I've come
to the conclusion that "stack" is a lot of difficulty for no benefit.

We originally created "stack" as an attempt to define a slightly
better version of the old "just put everything in 1/1" behavior.
Instead, it goes and finds the first *empty* slot, and puts everything
there; if you're not putting anything in 1/1, it'll have the same
behavior.  It's a little more complex than that to handle things with
a definite column position, though.

Having freshly written the algo, I just don't think it pulls its
weight.  It's still a terrible value, doing something you don't
actually want it to do.  It only exists to hopefully handle IE's
legacy content that depended on "auto-placing" things in 1/1, but it
might not even do that (if a page is currently positioning something
in 1/1 *and* depending on auto-placement to put more things there).  I
think we should just throw away "stack", add "none" with the behavior
that puts everything in 1/1 if its position is fully auto (and does a
simple "dense" packing for things that are auto in only one
dimension).  This would then prevent IE from having to make a
proprietary value for their old behavior, as we'd just match it.

Thoughts?

~TJ

Received on Tuesday, 16 September 2014 18:57:13 UTC