edgedb / imdbench

IMDBench — Realistic ORM benchmarking
https://edgedb.github.io/imdbench
Apache License 2.0
242 stars 24 forks source link

add go repack benchmark #13

Closed fmoor closed 4 years ago

fmoor commented 4 years ago

I got a little carried away here. Adding a new benchmark to the existing code was awkward and caused a lot of bloat because the query execution was mixed in with the result calculations. I separated the results calculations and aggregation from the logic related to preparing a client and executing a query. If this change is too large let me know.

fmoor commented 4 years ago

Looking at the benchmark results I might have a bug. Investigating :mag:

1st1 commented 4 years ago

Looking at the benchmark results I might have a bug. Investigating 🔍

Can you please share them? (A screenshot at least)

1st1 commented 4 years ago

If this change is too large let me know.

I like the way you refactored it, makes more sense now and IMO it's quite elegant.

fmoor commented 4 years ago

Can you please share them? (A screenshot at least)

The bug I mentioned was due to unpacking query results into the wrong struct type. That is fixed now, but the samples from go repack person are still missing values in the directed field. I'll look at that tomorrow.

These are the current results:

image

fmoor commented 4 years ago

I have convince myself that this is working now. Ready for review.