[Webkit-unassigned] [Bug 158281] New: SVGImage should report a more accurate more cost for its decoded data

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Wed Jun 1 16:03:39 PDT 2016


https://bugs.webkit.org/show_bug.cgi?id=158281

            Bug ID: 158281
           Summary: SVGImage should report a more accurate more cost for
                    its decoded data
    Classification: Unclassified
           Product: WebKit
           Version: WebKit Nightly Build
          Hardware: Unspecified
                OS: Unspecified
            Status: NEW
          Severity: Normal
          Priority: P2
         Component: SVG
          Assignee: webkit-unassigned at lists.webkit.org
          Reporter: sabouhallawa at apple.com
                CC: zimmermann at kde.org

In http://trac.webkit.org/changeset/201334, a simple calculation for the SVGImage memory cost was implemented. A more accurate but fast solution is required for better memory and resource caching utilization. Two solutions should be investigated:

1. A through navigation for the DOM and the render trees of the SVGImage accounting all the big memory chunks especially the vectors and the expensive cached graphics resources, e.g. the Path object. The down side of this approach is it is very costly and it's likely to become stale over time. 

2. Assume the SVGImage decoded data (the DOM and the render trees) is a ratio of the encoded data. This would require some heuristic to get this ratio offline. The ratio can be applied on the SVGImage level itself; e.g. sizeof_decoded_SVGImage = sizeof_encoded_SVGImage x decoding_ratio. Its down side is it is hard to get a good estimate for decoding_ratio. Also the result can be an inaccurate estimate especially if the SVG includes shared resources and they are referenced many times.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20160601/46c7f2d2/attachment.html>


More information about the webkit-unassigned mailing list