RE: SVG2: wrong definition for bounding box algorithm

Bjoern Hoehrmann <> wrote:

>> If `union` ignores that such a rectangle has no area then maybe, but 
>> that would be strange and unexpected, and I've certainly seen `union` 
>> implementations of geometry libraries that do not behave like that.
>> Note that there is the related open issue "Issue 11: Need to define 
>> what the union of rectangles with no area means."

Dirk replied:
>Well, why wouldn't

><path d="M100,100L100,100z"/>

>have a valid bounding box? You might not fill the area, but you can still
stroke it or add >a marker. So it is valid.
Interesting discussion. I think there is logic in both points of view. Does
a bounding box refer to the unions of nonzero areas or the unions of the
curves? A line of zero thickness has no area, but I think most would still
expect it to have a bounding box.

On the other hand, when one looks at the code generated by programs, there
may be extraneous points included in an SVG document. I assume that these
may represent places that the artist's hand shook or accidentally clicked.
They are, so far as I have seen, *never* part of the meaning of a document.
If a mathematician wants to draw a set which is the union of a line with a
point, the point will be made visible by giving it a non-zero radius. So, I
guess I am tempted to conclude that curves of zero measure (in the sense of
a zero-dimensional Hausdorff space [1]) should not contribute to the
bounding box since individual points do not contribute to the visual
semantics of a document.  



Received on Saturday, 19 April 2014 12:24:02 UTC