- From: Silvia Pfeiffer <silviapfeiffer1@gmail.com>
- Date: Thu, 7 Apr 2011 11:21:20 +1000
- To: Philip Jägenstedt <philipj@opera.com>
- Cc: public-media-fragment@w3.org
On Thu, Apr 7, 2011 at 2:12 AM, Philip Jägenstedt <philipj@opera.com> wrote: > On Wed, 06 Apr 2011 16:01:20 +0200, Jack Jansen <Jack.Jansen@cwi.nl> wrote: > >> >> On Apr 6, 2011, at 15:14 , Raphaël Troncy wrote: >> >>>> Yeah, I think we need to be more specific. My suggestion is to round >>>> up, but am curious what ppl think. >>> >>> +1 ... and added to list of items to discuss next week. >> >> >> ... And just after I pressed "send" I realised that "rounding up" isn't >> the right term. I think we should round in such a way that we enclose at >> least the specified region. Therefore, X and Y should be numerically rounded >> down, and width and height numerically up. > > The doesn't quite have the effect of enclosing the specified region, as > (10.5, 10.5, 20, 20) would get clamped to (10, 10, 20, 20), which misses 0.5 > px on the right and bottom. Similarly, (10.9, 10.9, 10.9, 10.9) would get > clamped to (10, 10, 11, 11), which misses 0.8 px on the right and bottom. > (The worst case with .999999 goes towards missing a whole row/column.) > > The formula for always enclosing the region would be: > > x' = floor(x) > y' = floor(y) > w' = ceil(x+w) - floor(x) > h' = ceil(y+h) - floor(y) Hmm, I see. Unless that is a problem with any existing implementations, I can now see an advantage to using (left, top, right, bottom) - which, incidentally, is also similar to how CSS works. Looking forward to the discussion! Cheers, Silvia.
Received on Thursday, 7 April 2011 01:22:07 UTC