westonpace commented on PR #33770:
URL: https://github.com/apache/arrow/pull/33770#issuecomment-1400855903
I think we should merge as-is since we don't want to let pending work block
current capabilities.
--
This is an automated message from the Apache Git Service.
To respond to the
westonpace commented on PR #33770:
URL: https://github.com/apache/arrow/pull/33770#issuecomment-1400796896
> Or should I be using ScanV2Options?
Yes, but it's not ready yet. Sorry if that was misleading. By "new" I
meant "the API we are working towards".
--
This is an automated
westonpace commented on PR #33770:
URL: https://github.com/apache/arrow/pull/33770#issuecomment-1400632307
> So I don't know why you need the projection expression at all, unless it
is aspirational/future-looking for some time when the projection can be pushed
down and handled by the file
westonpace commented on PR #33770:
URL: https://github.com/apache/arrow/pull/33770#issuecomment-1399270231
@nealrichardson Ok, I did some investigation.
First, the reason this is not being encountered from pyarrow:
The scanner options currently takes both a projected schema and
westonpace commented on PR #33770:
URL: https://github.com/apache/arrow/pull/33770#issuecomment-1399077722
> I thought this was just going to be deleting code from the R package:
instead of finding the top-level field names in the projection and sending them
in the ScanNode, I'd send the
westonpace commented on PR #33770:
URL: https://github.com/apache/arrow/pull/33770#issuecomment-1398740288
> I haven't run C++ unit tests in forever, so figured I'd get some feedback
before diving in there.
Sorry, I was thinking of R e2e tests. I would hope the C++ change is