Use benchmarks to drive development
In order to use the results of benchmarks to actually drive development, we need:
-
write one test that reflects user experience from soledad point of view (i.e. time for syncing 500 incoming messages). -
make that test run with legacy and blobs code. -
store test results. -
plot test results, parametrize by date/commit range. -
make the plots available in a public page (metrics.bitmask.net). -
do it all again for another test that reflects user experience from bitmask point of view (i.e. time for showing 500 incoming messages in inbox).
Edited by drebs