After OJ is enabled for AMS, the speed test randomly failed about fast_jsonapi HOT 6 OPEN

We may need to clean up and remove the json string serialization performance tests altogether or make them information only. The gem controls only the ruby hash creation time after that its just multi json and the chosen engine.

from fast_jsonapi.

@vovimayhem We can change the 25x criteria. if the feature really needs some performance degradation we will lower the criteria. I think we should always start by discussing if we can implement the feature without any material change to performance.

Here is an example of a sample discussion i enjoyed #49

from fast_jsonapi.

Agreed with @shishirmk . I love that the performance metrics are part of the process. PRs that do not impact performance can be approved quickly, but implementations that have a performance impact will at least raise a red flag to the maintainers. It may just require a minor tweak to fix the performance change, or we may be able to more tightly isolate the performance degradation to certain cases (e.g. a 10% degradation for polymorphic associations vs. a 10% degradation for all serialization).

from fast_jsonapi.

I understand - even praise - the performance criteria set up for this project. But maybe the "at least 25x" criteria is either a bit too much, or too inflexible for some scenarios that are actually hitting this limit.

I'm seeing (right now) master fail a couple of performance tests by 1x or less... leaving no headroom for adding new features :( - I'm taking a jab at #37 right now, and given master fails by 1x, my branch failing at around 2x on those same tests I feel it's a bummer.

from fast_jsonapi.

Agreed - see #64 :)

from fast_jsonapi.

This is still an issue after the release of version 1.1.0 so i think we need to tackle this and ensure that performance tests are stable and stop failing erratically

from fast_jsonapi.