On Sat, 2004-05-22 at 00:35, Bryce Harrington wrote:
There should already be bugs/RFEs for both. The coordinate system fix is probably major enough that we need to find space on the roadmap for it.
If you can send me a list of SVG-compliance items like this that need to be planned into the roadmap, I'll take care of fitting everything in.
In order of decreasing nastiness (I think):
* [SVG] fix the coordinate system -- this will require modification of a lot of wide-ranging code which we still don't understand. The coordinate system cleanups performed by Peter et al will help a little, but I suspect it's still going to be rather brutal.
I think it's best we plan another straight bugfix release before the release which attacks this, so people will have a recent stable release to use while any lingering issues get shaken out.
* [SVG] Redo the viewBox UI, and change how the viewBox on the root <svg> node is displayed (for nested <svg> elements, the current behavior is correct and desirable) -- one of the SVG guys suggested we add an extra outlined box denoting the viewbox.
* [XML] Rework the Repr -> SPObject lookup to use something other than IDs. I think mapping by pointers should be good enough.
The problem is currently that we blindly mash an id= attribute on all elements -- while that is valid in the SVG, inkscape, and sodipodi namespaces, other namespaces and their associated schemas may have different IDREF attributes or none at all. Under some circumstances, that can be very destructive to e.g. embedded RDF fragments.
We'll still need ids for things like <use>, obviously, but we should generate such ids only as needed, and only in namespaces in which we know what the IDREF attributes are.
-mental