As discussed on the WebPerf WG call [1], the experimental `transferSize` implementation [2] should align to the spec [3] and well as the outstanding relevant PR [4], and expose: * The encodedBodySize + 300 for resources that were fetched over the network * 0 for resources that were fetched from the cache * 300 for resources that were successfully revalidated over the network, and their body was then delivered from cache. [1] https://docs.google.com/document/d/10dz_7QM5XCNsGeI63R864lF9gFqlqQD37B4q8Q46LMM/edit?pli=1#heading=h.gkilfdieaqp [2] https://github.com/WebKit/WebKit/blob/b5cd1015460639460f55ae4b05bc90ff4ea5125e/Source/WebCore/page/PerformanceResourceTiming.cpp#L246 [3] https://w3c.github.io/resource-timing/#dom-performanceresourcetiming-transfersize [4] https://github.com/w3c/resource-timing/pull/272
<rdar://problem/80074642>