According to tests double type is unnecessary. SVG Filters use float type everywhere so we shoud change double to float.
Created attachment 152751 [details] proposed patch
What is the expected benefit of this change? Is it matching other code, or something else?
Opengl ES 2.0, ARM NEON, OpenCL don't support double type.
Please elaborate. This patch changes C++ code, not OpenGL or OpenCL.
For this filter doesn't need double precision and it confused me when I implemented the opencl/gl version. The goal is to uniform the codes.
Thank you for the explanation. I'll leave final review to someone better familiar with this code, and with performance characteristics of double vs. float on modern processors.
Comment on attachment 152751 [details] proposed patch Looks okay to me - all tests still pass, so we're not affected by precision issues, so rs=me.
Comment on attachment 152751 [details] proposed patch Clearing flags on attachment: 152751 Committed r123288: <http://trac.webkit.org/changeset/123288>
All reviewed patches have been landed. Closing bug.