Håkon V. Treider
Håkon V. Treider
Not sure if this is the best way (stencil should maybe be used?). I don't get any performance gains unless I cheat and run in parallel. With core count 4...
> Are you using Anaconda's NumPy packages? Those link to MKL, which does a pretty good job speeding up a number of ufuncs. Nope, using standard pip install (openblas afaik)....
Just using `--proto3` solved it for me
List of affected: 1. `AggregateResult(dict)` 2. `AggregateResultItem(dict)` 3. `AssetAggregate(dict)` 4. `BoundingBox3D(dict)` 5. `DataSetAggregate(dict)` 6. `EndTimeFilter(dict)` 7. `ExtractionPipelineContact(dict)` 8. `FileAggregate(dict)` 9. `GeoLocation(dict)` 10. `GeoLocationFilter(dict)` 11. `Geometry(dict)` 12. `GeometryFilter(dict)` 13. `Label(dict)` 14....
> I took a look at the affected classes. And only around a quarter of them have `dump()` method. As I understand, if the dump method is absent there is...
To be solved by #1452
@andeplane What about rather always importing them explicitly? I.e. change all to `from ... import And, Equals`
[Treider-Bot] PR seems stale, closing in 2 weeks 🤓
> Describe the solution you'd like > > 2. Perhaps add an expand_metadata argument to the to_pandas method allowing users to control behavior. Coming in the next major version: #1453
# Interested in feedback on how to make the class hierarchy better! - Type hints still feel lacking in many places, should we add `typing_extensions ^4` to get `Self` for...