[cssom][css-display][css-cascade] revisiting "default display"

At risk of beating a dead horse, allow me to describe two use cases, and
how CSS Display Level 3 (and CSSOM and CSS Cascading) attempts to address
each one:

1. A javascript library wants to be able to "hide" (set display:none) a
"non-hidden" (has a display other than none) element and then "show"
(return to the display value before setting it to none) it.  This is the
main use case covered by Display Module Level 3. By providing a separate
"box" property, a javascript library can set that property to "none" to
perform a "hide", and then set that property to "normal" to perform a
"show". No loss of information about whether the element was inline or
block. Great!

2. A bit more insidiously, a javascript library wants to "show" a div that
is already "hidden" through an element selector in the author style sheet:
div {display:none;}. Can Display Level 3 address this? The current spec
says:

The computed value of box <http://dev.w3.org/csswg/css-display-3/#box> is
the specified value,
unlessdisplay-outside<http://dev.w3.org/csswg/css-display-3/#display-outside>
has
the value none <http://dev.w3.org/csswg/css-display-3/#display-outside-none>,
in which case the computed value of
box<http://dev.w3.org/csswg/css-display-3/#box>
 is none <http://dev.w3.org/csswg/css-display-3/#display-outside-none> as
well.

I believe this prevents a JS library like jQuery from overcoming authors
who use element selectors to define display:none. This may not seem like a
big use case, but happens often enough. We're left with the situation where
Display 3 still needs a mechanism by which an element can be reverted to
its "default display". Now, we can add a value to display-outside like
default that accomplishes this, but now its starting to sound very familiar
to the generic "default" keyword that was just dropped from cascade level 3.

To be clear, currently a javascript library must append an iframe to the
page to get at the "default value" of display for the div. CSSOM had at one
point defined an interface created by dbaron called
getDefaultComputedStyle() to essentially emulate the iframe hack. In
conjunction, Cascading 3 at one point had a "default value" to define what
value gDCS returns.

I know I raised this issue and subsequently acquiesced when Cascading Level
3 was still LC, but I find myself still needing a solution to this problem
not yet covered by any spec.

Mike Sherov
Chief Technologist
SNAP Interactive, Inc. | Ticker: STVI
http://snap-interactive.com | http://ayi.com