Research on ISSUE-2346

I had a look through the threads in ISSUE-2346 for ACTION-2908.

For a):
I propose that we define the rounding method for 'filterRes' to be the  
same as for 'seed', i.e truncation (round-towards-zero).

For b):
I think it's probably better to remove some parts here.

In my experience 'filterRes' is usually best if left alone, since having a  
predetermined size of the buffers means that the implementation can't  
allocate smaller buffers for when the computed filter region is small for  
example. However, it can be used for some specific cases to degrade /  
speed up the filter processing. The content becomes dependent on the  
device and renderingsize, or in other words becomes less portable.


To resolve this issue, I propose that we replace this old text[1]:

[[
   This attribute takes the form x-pixels [y-pixels], and indicates the  
width and height of the intermediate images in pixels.
   If not provided, then a reasonable default resolution appropriate for  
the target device will be used.
   (For displays, an appropriate display resolution, preferably the current  
display's pixel resolution, is the default.
    For printing, an appropriate common printer resolution, such as 400dpi,  
is the default.)
]]

with the this new text:

[[
   This attribute takes the form x-pixels [y-pixels], and indicates the  
width and height of the intermediate images in pixels.
   If not provided, then the UA will use reasonable values to produce a  
high-quality result on the output device.

   If a non-integer value for 'filterRes' is specified, the value must be  
truncated, i.e rounded to the closest integer value towards zero.
]]

Cheers
/Erik

[1] http://www.w3.org/TR/SVG11/filters.html#FilterEffectsRegion

-- 
Erik Dahlstrom, Core Technology Developer, Opera Software
Co-Chair, W3C SVG Working Group
Personal blog: http://my.opera.com/macdev_ed

Received on Monday, 29 November 2010 11:37:50 UTC