guiyanakuang commented on pull request #810:
URL: https://github.com/apache/orc/pull/810#issuecomment-893101538
--
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 unsubscri
guiyanakuang commented on pull request #810:
URL: https://github.com/apache/orc/pull/810#issuecomment-893109802
[ORC Specification v1](https://orc.apache.org/specification/ORCv1/)
![image](https://user-images.githubusercontent.com/4069905/128279409-76fa1e1c-beae-49b5-a440-aa4916c883f4.pn
guiyanakuang commented on pull request #810:
URL: https://github.com/apache/orc/pull/810#issuecomment-893101538
> How many users metadata do you have, @guiyanakuang ?
Usually around 10.The largest have thousands of this order of magnitude. It
may be abusing this feature, but so far t
guiyanakuang commented on pull request #810:
URL: https://github.com/apache/orc/pull/810#issuecomment-892405840
I think I misunderstood, the user is also allowed to actively call methods
to get the metadata, so that there are two data structures. Perhaps an internal
method could be added t
guiyanakuang commented on pull request #810:
URL: https://github.com/apache/orc/pull/810#issuecomment-892400814
The metadataCache is only initialised when the metadata is needed for
compatibility checking or merging. In fact the reader is quickly cleaned up and
only survives in the merge c
guiyanakuang commented on pull request #810:
URL: https://github.com/apache/orc/pull/810#issuecomment-892378836
> @guiyanakuang . Technically, this looks reasonable because Java
`WriterImpl` ensures that there is no duplicated keys at least Java
layer(writer and reader)
>
> ```java