thats 58% of the round-trip
On Mon, Jul 29, 2013 at 2:12 PM, Michal Mocny <[email protected]> wrote: > Just to be clear, thats 58% of *total* exec round trip time (580ms) , or > 58% of only the native bits? > > If its over half of the total, thats ridiculous. > > -Michal > > > On Mon, Jul 29, 2013 at 11:36 AM, David Kemp <[email protected]> wrote: > > > I have been doing some testing on the bridge to see how the current > > implementation stands up. > > > > For large data going to native, the highest single time impact is the > > conversion of the string rawArgs to a JSONArray. > > > > Currently Echo, FileUtils, FileTransfer are likely to be impacted by > this. > > > > The plugin execute is called by the bridge with String rawArgs, and the > > signature of these plugins (and others) calls for JSONArray args. This > > causes a non-obvious invocation of JSONArray(string) that isn't > > particularly fast. > > > > Testing on a Nexus 7 > > for a 1MB echo call (total round trip 580ms) > > 58% - conversion from json string to JSONArray (in Native) > > 28% - js preparation (Process any ArrayBuffers in the args into a > string.) > > > > more to follow... > > >
