« Gran Paradiso Alpha 1 and SVG | Main | SVG Priorities in Firefox 3 »
February 7, 2007
Gran Paradiso Alpha 2 and SVG
Gecko 1.9a2 has been released. There's been quite a few SVG enhancements since a1, but unfortunately a regression found right before the tree froze for a2 means that SVG files with a certain construction won't be rendered (bug 369502). If your content works with that bug, but you see other issues, please file bugs with testcases.
One of the largest changes is that the version of cairo we're using has been updated to 1.3.12. This includes a number of performance enhancements, most noticably a new tesselator which should help with complex geometry. Gradients should also be faster, though there is a visual artifact bug with reflect mode linear gradients.
These are the highlights of the changes since 1.9a1:
- Text will move when x/y attributes are changed.
- Default mode of <feMorphology> is erode.
- <image> will use less memory, dramatically so in the case of the same image being referenced multiple times, and should be much faster rendering.
- <feBlend>, <feComposite>, <feFlood>, and <feTurbulence> implemented.
- Removing text node will cause display to change.
- Nonstandard DOM method SVGTransformList.getConsolidationMatrix() removed.
- <svg> live to length changes.
- Group opacity calculation avoided when possible. Should speed up the case where "opacity" is accidentally used instead of "fill-opacity" or "stroke-opacity".
- Having minimum font size set in browser won't mess up SVG text now.
- Fallback for paint servers implemented.
- SVGPathSegList fully implemented.
- <a> now implemented natively, should be more specification complient.
- Filters can operate in "linearRGB" or "sRGB" as specified by the document.
- <switch> is now SVGTransformable.
Posted by tor at February 7, 2007 11:22 PM
Comments
Are you guys aware that the neat dynamic text-on-a-path testcases in Bug 282579 don't work anymore?
Posted by: DigDug at February 8, 2007 1:30 PM
Thanks for pointing out that those were broken. The issues appear to be a combination of problems; some recent regressions that we now have patches in hand for, and an older one caused by an optimization to pathseglist creation which I'm investigating now.
Posted by: tor at February 13, 2007 4:24 PM
great !
Posted by: Antoine Magnier at February 27, 2007 3:42 PM
i don't think Adobe want it be :(
but we have freedom!!
Posted by: at March 7, 2007 2:20 PM
tank you mozilla
Posted by: tsabat at June 26, 2007 1:57 PM