RE: [gcpm] border-parts

Also sprach Alex Mogilevsky:

 > - It looks pretty cool now. The definition is simple but very powerful!

Indeed.

 > - I wonder if there is another word other than 'parts' that would
 > better describe it? Maybe 'border-fragments'? Or 'border-pattern'
 > (which is what it actually is)?

I like border-pattern. It's a little longer than border-parts, but it
avoids the plural.

Any other opinions on this?

 > - Percentages - I think they should be always of width (logical
 > with) -- then it is consistent with border-width, and it is
 > possible to define percentage-sized dashed border.

It should probably be the same as for the margin properties:

  Percentages: 	width* of containing block
  *) if the containing block is horizontal, otherwise the height

  http://www.w3.org/TR/css3-box/#margins

 > - making it more intuitive -- maybe syntax like
 > 
 >         border-parts: on(4em) off(1em) on(4em);
 > 
 > would make it more intuitive, but I find current one more practical...

This could work. Then we could do:

  border-pattern: on(4em) off(1em) off(1fr) on(4em)

instead of the more cryptic:

  border-pattern: 4em 1em 0 1fr 4em

But this is still shorter, and quite readable actually, so I think we
should keep it.

 > One more thing... I think we talked about defining the parts
 > literally as clip region that erases portions of the border. The
 > advantage of using that definition is that it tells UA precisely
 > what is expected, especially in complex cases like elliptic
 > corners, insets, etc. However if there is a more intelligent way of
 > handling it, that definition would not allow it... I think it is
 > still worth mentioning as suggested implementation.

These clip regions would be rectangular, right?

-h&kon
              Håkon Wium Lie                          CTO °þe®ª
howcome@opera.com                  http://people.opera.com/howcome