gianm commented on code in PR #19845:
URL: https://github.com/apache/druid/pull/19845#discussion_r3837639962


##########
server/src/main/java/org/apache/druid/catalog/model/table/DatasourceDefn.java:
##########
@@ -107,18 +106,6 @@ public void validate(ResolvedTable table)
     super.validate(table);
     final DatasourceBaseTableMetadata baseTable = 
table.decodeProperty(BASE_TABLE_PROPERTY);
     if (baseTable != null) {

Review Comment:
   > You're right that CompactionTask will skip downloading the segments and 
treats baseTable spec if defined in an inline config as authority if it is set, 
my current thinking is that this is the equivalent of a catalog sealed table.
   
   Are you saying that the CompactionTask check I mentioned is only going to 
fire for inline specs, and for inline specs, `baseTable` still requires the 
table to basically be treated as sealed? (i.e. if extra columns exist, they 
will be dropped.)
   
   I suppose that's fine, although it feels like it should be adjusted at some 
point to be more consistent with the catalog behavior. Maybe there should be a 
`sealed` parameter for the inline specs too. Anyway, I suppose it's ok to 
address this in a future PR.
   
   I believe the important thing for this PR would be to make sure that if/when 
compaction uses a `baseTable` from the _catalog_, that it respects the `sealed` 
parameter or lack thereof. (If `sealed` is not set, then compaction should go 
figure out the full set of columns.) I think you are saying that compaction 
can't currently pull `baseTable` from the catalog. If that's true then I think 
it's fine to deal with integrating that stuff in a future PR. When it _is_ 
integrated, `sealed` should be respected.



-- 
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]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to