Re: Any new comments on the non-uniform scaling patch?

[Lars Clausen <lars raeder dk>]

On Tue, 2007-12-04 at 12:28 +0100, Marcel Toele wrote:
> Lars/any dev with svn acces/anyone else,
>
> Can you please comment on the updated information about the
> non-uniform scaling patch 
> (I've added extra information via Replies to that thread)

Sorry it took a bit, I've again been busier than I hoped for.  I got the
patch to work as intended, and can now start to look at problems with
it.  

A major but fixable problem is that the bounding box does not get
updated correctly when a sub-shape is large enough to be outside its
super-shape, leading to dirt on the diagram.  However, some of your
previous comments seem to indicate that you expect sub-shapes to always
be within the super-shape.  If that is the case, the super-shape should
limit its resize if it would put the sub-shape outside it.

The sub-shape in the example that's anchored at the lower right doesn't
appear, but that's a bug in the SVG path for it.  Once I copied the
first star's paths, it turned out fine, and the bottom/left anchors work
fine.

I'm not convinced that the default_scale attribute is the right way to
go about handling the scaling issues.  It means that whereas the
super-shape ignores the absolute scale and just uses the coordinates as
relative coordinates in a 2x2 cm space, the sub-shape actually uses the
units of the SVG, modified by the default_scale.  This is confusing: The
coordinates of the super-shape and the sub-shape are now interpreted in
different ways, one of which is controlled by a "default_scale"
attribute whose meaning is not clear from context.  If the default_scale
was on the main shape, it would be less confusing and at the same time
close a long-standing bug.

Another problem with the default_scale is that it's defined relative to
something that's not in the shape definition, namely the unit system.
mm might be default in many locals (I frankly don't know, and would
rather not have to remember), but how would the shape be rendered in the
US, where inches or suchlike are the default?  As I see it, the
coordinates for the super-shape and the sub-shape should have the same
meaning, or else editing the shape becomes a trap.  Rather than the
default_scale attribute we should, IMNSHO, have either a unit definition
or a bounding-box definition with units. 

Am I right to think that there is no such thing as a sub-sub-shape?
When I try, the sub-sub-shape appears, but not different from when it's
a sub-shape.

I'm not included to apply the patch until we have agreement on the
default_scale matter.

-Lars