guiyanakuang commented on pull request #952: URL: https://github.com/apache/orc/pull/952#issuecomment-964903950
> > > @guiyanakuang I am wondering whether the selected vector only applies to the root struct-typed column vector. And I am curious if this patch is a part of some other big work items? > > > > > > @wgtmac Yes, the selected marker is at the row level. It will only work on the root. The purpose of this pr is to resolve [issue](https://issues.apache.org/jira/projects/ORC/issues/ORC-1004?filter=allopenissues). This feature helps me to write the results to the ORC file by filtering, which is useful for creating materialized views. It just happens to be useful for my current work, so I implemented it > > As the selection vector only applies to the root struct type, is it possible to use null buffer of the root StructColumnVector (which is always not null for now)? We can ignore writing rows of null data there. This is much simpler but hacky. Sorry I didn't answer accurately, but in fact the root may also be a single column primitive type. Also the final VectorizedRowBatch obtained by filter query, the unselected rows may not be null. StructColumnVector seems to be written in such a way that it only appears when the sub-column type is struct. -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
