danielcweeks commented on a change in pull request #1844:
URL: https://github.com/apache/iceberg/pull/1844#discussion_r545335609
##########
File path: api/src/main/java/org/apache/iceberg/io/FileIO.java
##########
@@ -29,7 +29,7 @@
* must be serializable because various clients of Spark tables may initialize
this once and pass
* it off to a separate module that would then interact with the streams.
*/
-public interface FileIO extends Serializable {
+public interface FileIO extends Serializable, CatalogConfigurable {
Review comment:
I just want to be sure that we understand how the properties are getting
propagated to the FileIO. Right now, it seems that we're loading it via the
CatalogUtils, but only in the glue catalog, so it seems somewhat specific at
this point. Since the glue catalog can just load the properties and use the
S3FileIO constructor, it seems like we should pull it out from the PR and
revisit with a broader change address the default dynamic loading.
I agree that it might be a good addition because if you dynamically load
FileIO right now, I don't believe you can set custom AwsProperties.
----------------------------------------------------------------
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
