On 7 Feb 2011, at 17:51, Richard Fairhurst wrote:
> Some interesting reading today!
>
> Rather than going through stuff point-by-point I'd just make one general
> observation, which is: simple is good; approachable is good. Ok, that was two
> general observations.
>
> If a newbie who knows some (web) CSS can write:
>
> way[highway=motorway] { width: 10; color: #0000FF; }
> way[highway=trunk] { width: 8; color: #00FF00; }
> way[highway=primary] { width: 8; color: #FF0000; }
>
> and click "refresh", then see the map appear as they expect, then that's
> great. We have a convert.
>
> The "implement layers through eval and z-index" stuff is clever and
> architecturally neat, but doesn't pass the above test. It would be kind of
> like removing <p>, <ul>, <li> and everything from HTML, simply leaving <div>
> and <span>. Architecturally beautiful and utterly logical (after all, they
> can all be specified in CSS), but a PITA to work with.
>
> Similarly for the colour stuff. Having multiply operations and all of that
> would be great. But it shouldn't be at the cost of rejecting a universally
> known, though perhaps slightly imperfect, standard.
I don't really see how the layers proposal would change the above example (or
any simple example), with a good quality default style sheet (that maps layers
into z-indexes nicely). Of note one of the most powerful things about css is
that you get to do things like go "see that list thing in the html, I don't
want it to be a list". The same thing should be true of MapCSS – it should be
possible to take the layers specified in the tags, and entirely rearrange them.
A practical example to justify why this is something we'd want: when selecting
ways in an editor it would be useful to make those ways appear on top of
un-selected ways, with highlit casings (Potlatch does exactly this in fact).
Being able to simply specify
way :selected
{
casing-z-index: eval(100 + prop(casing-z-index));
z-index: eval(100 + prop(casing-z-index));
}
(making an assumption that 100 is higher than any other way).
Thanks
Tom Davie
_______________________________________________
Mapcss mailing list
[email protected]
http://lists.openstreetmap.org/listinfo/mapcss
