As part of the current WebKit stabilization effort, we'd like to disable some of the more experimental SVG features, including: - SVG in images - Animation - Filters - <use> - <foreignObject> I would suggest adding a single define SVG_ENABLE_EXPERIMENTAL_FEATURES, to guard them all, so it is still possible to test them for those who want to.
<rdar://problem/5021698>
Once this is resolved, bugs with these features will no longer be P1.
How experimental is <use> support? A number of SVGs use it and it would be a shame to have it miss the boat.
Nikolas's assesment of the situation when it comes to <use> in http://lists.macosforge.org/pipermail/webkit-dev/2007-February/001683.html leads me to think that <use> is quite testable and supported well enough to make the cut as a stable feature. Given the high value of this feature and how widely used it is in SVG content out there, I sincerely hope it's not marked as experimental.
My plan is to disable <use> for now with the other bleeding-edge features. If someone wants to do extra testing for it, we can reconsider and flip the switch back at some point.
Created attachment 13378 [details] patch to add ENABLE_SVG_EXPERIMENTAL_FEATURES define This patch doesn't include the changes to numerous SVG test cases that result from it, I would land those with the patch. However, we may want a buildbot with experimental features on so these don't stop getting regression testing.
The patch also reworks how optional features in general are handled, so now the code says things like: #if ENABLE(XPATH) instead of: #ifdef XPATH_SUPPORT
Comment on attachment 13378 [details] patch to add ENABLE_SVG_EXPERIMENTAL_FEATURES define - print "#ifdef SVG_SUPPORT\n"; + print "#if EABLE(SVG)\n"; should be ENABLE. The SVG filters in platform/graphics/svg should not get compiled in. With these two changes it's r=me.
Landed (adjusted for some of Lars's comments).