Re: [1.2T-LC] inverse and constrained transformations (ISSUE-2073)

Anthony Grasso:
> Hi Dr. Hoffmann,
>
Hello ;o)

> Excellent find in the specification! 

It is old and I already noted this previously a longer time ago, but 
there was not much interest to care about it or to test it. At least 
I have related tests about this issue in my own test suite for 
more than two years ;o)

Note, that Opera 9.5 manages both critical situations with determinant
0 and indefinite (due to a possible skew transformation of for example
90 degree) and has only minor problems in relation with vector-effect
non-scaling-stroke (instead of a constrained transformation).

I think, this indicates already, that Opera uses another (more clever?)
method to get the effect of the constrained transformation anyway, 
obviously without using the inverse matrix and avoiding the problem.
On the other hand, the problems of Opera with vector-effect 
non-scaling-stroke indicates that it maybe could be even more clever -
especially because for non-scaling-stroke an inversion is not even
mentioned ;o)

> The SVG Working Group discussed this 
> issue and we concluded that we should add wording that warns authors of the
> problem.
>
> As a result I was given ACTION-2284 [1] to add the wording to the
> specification. The new wording can be found in "The TransformRef value" [2]
> section of the specification.

Can you point out more precisely the position of the change?
I looked into the section twice and could not find a change from
the working draft to the editors draft.


Olaf

Received on Saturday, 11 October 2008 17:25:23 UTC