At the moment we seem to be creating a new DataBuffer for every path segment. DataBuffers are not very lightweight objects and we shouldn't create so many. In Zombotron 2 for instance, we parse nearly 283,135 such segments.
Yeah, that is the single biggest cause of slow parsing. I'll overhaul shape parsing after I land bug 1035170.
Status: NEW → ASSIGNED
Depends on: 1035170
Shape parsing is a bottleneck for fonts and vector graphics.
Priority: -- → P3
Summary: User fewer DataBuffers when parsing PathSegments. → Use fewer DataBuffers when parsing PathSegments
Parsing is much less of a bottleneck overall, but we should at least spend some effort on optimizing this before launch.
4 years ago
Status: ASSIGNED → RESOLVED
Closed: 4 years ago
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.