Expand Arrow reading support

[Kopilov]

Arrow reading test with more data examples (generation is presented at https://github.com/Kopilov/arrow_example).

Read UInt-s without overflow, read date and time as date and time.

NullabilityOptions pattern.

[koperagen]

This test fails. Can you check? We use it as a sample for docs
https://github.com/Kotlin/dataframe/blob/master/tests/src/test/kotlin/org/jetbrains/kotlinx/dataframe/samples/api/Read.kt#L79

Also, in public class ArrowFeather : SupportedFormat readDataFrame functions should set NullabilityOptions.Widening as an argument

[Kopilov]

This test fails. Can you check? We use it as a sample for docs
https://github.com/Kotlin/dataframe/blob/master/tests/src/test/kotlin/org/jetbrains/kotlinx/dataframe/samples/api/Read.kt#L79

This is because I removed file ./tests/src/test/resources/data-arrow_2.0.0_uncompressed.feather (that was originally copied to ./dataframe-arrow/src/test/resources/data-arrow_2.0.0_uncompressed.feather). Is using only one of them possible or we can only restore the copy?

[Kopilov]

Also, in public class ArrowFeather : SupportedFormat readDataFrame functions should set NullabilityOptions.Widening as an argument

Do I understand correctly that we should use Widening in SupportedFormat readDataFrame and Infer (formerly Keeping) as default in DataFrame.Companion.readArrowFeather methods?

[koperagen]

Also, in public class ArrowFeather : SupportedFormat readDataFrame functions should set NullabilityOptions.Widening as an argument

Do I understand correctly that we should use Widening in SupportedFormat readDataFrame and Infer (formerly Keeping) as default in DataFrame.Companion.readArrowFeather methods?

Yes

[koperagen]

This test fails. Can you check? We use it as a sample for docs
https://github.com/Kotlin/dataframe/blob/master/tests/src/test/kotlin/org/jetbrains/kotlinx/dataframe/samples/api/Read.kt#L79

This is because I removed file ./tests/src/test/resources/data-arrow_2.0.0_uncompressed.feather (that was originally copied to ./dataframe-arrow/src/test/resources/data-arrow_2.0.0_uncompressed.feather). Is using only one of them possible or we can only restore the copy?

I think a copy won't harm

[koperagen]

@Kopilov Thank you!

[Kopilov]

@koperagen you are welcome!

Our next step is to make saving :-)