gortiz opened a new pull request, #10715: URL: https://github.com/apache/pinot/pull/10715
This is a #10713 bugfix. As explained there, the problem occurs when user specifies that the column should not have a dictionary but that column is also included as a sorted column. In that case, pre `index-spi` we just ignored the `noDictionaryColumns`, but the newer code honored the config. An alternative is to read the dictionary even if the config says it should be disabled. But I think is more difficult to make mistakes if we actually introduce this (probably undesired) special case when reading the config in the old syntax. Therefore this PR modifies the deserializer used by DictionaryIndexType to add the following logic: - If the user disabled the dictionary using the older config BUT set the same column as the sorted column, the DictionaryConfig for that column is considered enabled. - If the user disabled the dictionary using the new syntax BUT set the same column as a sorted column, an error is thrown saying that the sorted column must be dictionary encoded. By doing that we preserve backward compatibility while we encourage people to use a coherent TableConfig once they migrate to the new configuration syntax -- 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: commits-unsubscr...@pinot.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@pinot.apache.org For additional commands, e-mail: commits-h...@pinot.apache.org
