Re: Stroke-miterlimit handling

Am Montag, 13. Mai 2013 schrieb Steve Schafer:
> On Fri, 10 May 2013 17:48:05 +0100, you wrote:
> >according to my precision animation test, firefox seems to have only an
> >accuracy problem, available in all versions, from 2 up to 20.
>
> I don't think that that can explain what I'm seeing in Firefox 20.0.1 on
> Windows 7. It's not a question of at what angle the miter limit
> threshold takes effect; rather, the _location_ of the clip point is
> different. In particular, as the join angle is animated through the
> threshold, the "correct" behavior is for the point to be suddenly
> clipped back, as if as if the stroke-linejoin property were changed from
> "miter" to "bevel." What Firefox does is different: the line is
> certainly clipped, but much further out, _not_ at the place where it
> would have been clipped with stroke-linejoin="miter." An animation of
> the process is continuous and smooth, with no discontinuous jumps.
>
> -Steve Schafer

Interesting, Firefox 20.0.1 von Debian Linux seems to switch in my
animation only between miter and bevel due to accuracy problems,
no change of the position of the clipping (my test path consists of a
horizontal segment and another straight segment under a variable angle
to the horizontal segment, therefore it is simple to see the position, 
where a possible miter has to start - this is compared with computed
paths for the outline of the stroke including the miter).  
Maybe they use another graphical library for windows as for other
operating systems? 
For WebKit it is known, that they use different (graphical) libraries
respectively the WebKit library can depend on quite different libraries, 
therefore always a surprise, what you will get ;o)

Olaf

Received on Monday, 13 May 2013 13:54:31 UTC