Re: [css-logical-props] Is this spec ready to ship?

On Mon, Dec 19, 2016 at 10:49:39PM +1100, Peter Moulder wrote:
> The only remaining issues for the size,padding,border,margin properties
> that I'm currently aware of are:
> 
>   - #691: whether it's computed or used 'direction', and hence whether
>     the computed value of 'text-orientation' is relevant.
> 
>     (Marked as closed, but see recent comments on the bug.)

Test case:
  http://www.users.on.net/~pjrm/css-tests/logical-props/upright.html

Note that the issue doesn't affect any of the UAs I tested, because they
haven't finished implementing text-orientation:upright: used 'direction'
was always simply equal to computed 'direction' in the UAs I tested (based
on text overflow direction).  Naturally, this issue isn't be a blocker for
any UA where used 'direction' is simply equal to computed 'direction'.
[Though incorrect rendering of text-orientation:upright might be
 considered a blocker for shipping that feature.])

>   - #800: meaning of percentages on logical props.
> 
>   - Relatedly, meaning of 'inherit' on logical props.

Test cases:
  http://www.users.on.net/~pjrm/css-tests/logical-props/percent.html
  http://www.users.on.net/~pjrm/css-tests/logical-props/inherit.html

I've only tested one old version each of Chromium & Firefox;
as expected, behaviour in each was as if the same value had been specified on
the corresponding physical property.

I haven't considered whether there's an authoring advantage to some other
approach, but since it's already been resolved to let shippedness trump
longer-term authoring value for more important decisions, I propose that the
spec say this explicitly (and that it say that this is true for all specified
values for logical property names).


The directory

   http://www.users.on.net/~pjrm/css-tests/logical-props/generated/

generated by the script

   http://www.users.on.net/~pjrm/css-tests/logical-props/gen.py

might also be of some interest in decisions of shipping: tests of border and
*size properties.  (Though it doesn't include any tests of margin or padding
properties at present.)  All pass in the old versions of Firefox & Chromium
that I tested.

pjrm.

Received on Thursday, 22 December 2016 06:10:58 UTC