emkornfield commented on pull request #165:
URL: https://github.com/apache/parquet-format/pull/165#issuecomment-767097727
@nevi-me I don't think we should be imposing arrow's modeling of interval
type on parquet. The existing interval type seems reasonable in parquet. I
think there are three cases to consider:
1. Arrow writing to parquet (this is well defined without data loss).
2. Arrow Reading from a parquet file that was written with arrow. In this
case, the additional written arrow schema should be sufficient for decoding.
3. Reading interval written from another implementation. In this case, I
think we probably want a user configurable options in arrow, as there can be a
few options:
1. Read as Duration
2. Read as struct of both interval types.
3. Read assuming as of the two interval types and truncate/error out data
that doesn't fit in the type.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
