Closed d-uzlov closed 2 years ago
@anhappdev please help on this one. do some initial investigation
I would store the result.json
file of each benchmark run on disk with a timestamp. In the history screen, the app shows a list of those result files, and on tap of one result file, the app show the content of the result file.
When working on https://github.com/mlcommons/mobile_app_open/issues/251 I have modified the way we handle result.json
a bit. I fixed serialization and added few new fields that are required for uploading (I think these fields will also be useful for history).
I'll extract it into a separate PR today so you can use it for result history.
@d-uzlov Would you take over this issue? I think with the tasks you are current doing (implementation of new result spec), this issue would fit better in your workflow and design.
@anhappdev sure.
@d-uzlov Great. Thanks.
I have created 2 separate issues for backend changes and for UI changes.
Currently the app only shows last result. When you rerun any set of benchmarks, results of the previous run are erased. We need to add a new UI page that will show list of previous results and allow to inspect them.
This feature must be offline. We can technically upload result into Firebase, and then fetch them to show in history but sometimes this can be undesirable, so we must keep results locally.
Initial discussion was here: https://github.com/mlcommons/mobile_app_open/issues/235
Steps: