CSV reader example for the cpp client #1455
Conversation
|
Fixes #892 |
jcferretti
changed the title
| final ArrowType.Timestamp timestampType = (ArrowType.Timestamp) arrowType; | ||
| final String tz = timestampType.getTimezone(); | ||
| final TimeUnit timestampUnit = timestampType.getUnit(); | ||
| if (tz == null || "UTC".equals(tz)) { |
There was a problem hiding this comment.
This is what I was worried about on the call, and makes me happy enough to ignore the rest for now, a good error indicating that we are just punting on anything else interesting.
| throw GrpcUtil.statusRuntimeException(Code.INVALID_ARGUMENT, exMsg + | ||
| " of floatingPointType(Precision=" + floatingPointType.getPrecision().toString() + ")"); | ||
| } | ||
| case Utf8: |
There was a problem hiding this comment.
Also LargeUtf8 here.
And probably Null if we're adding to this again, which is easy, it sends nothing, all nulls.
And possibly binary/largebinary/fixedsizebinary would map easily to byte[] or DbByteArray.
Seems like if we're handling near matches, we should address most of the things - Interval and Time are mostly "shaped like" and used like Duration, for example, and any Decimal can be made to work in BigDecimal (but not vice versa - then again most of these others including TImestamp won't behave vice versa either).
| return double.class; | ||
| case HALF: | ||
| default: | ||
| throw GrpcUtil.statusRuntimeException(Code.INVALID_ARGUMENT, exMsg + |
There was a problem hiding this comment.
or possibly just upconvert to float - i mean that's what we're doing with less precise timestamps that we don't support...
No description provided.