compasses commented on code in PR #11579:
URL: https://github.com/apache/doris/pull/11579#discussion_r1052989488
##########
fe/fe-core/src/main/java/org/apache/doris/analysis/IndexDef.java:
##########
@@ -37,6 +37,11 @@ public class IndexDef {
private String comment;
private Map<String, String> properties;
+ public static final String NGRAM_SIZE_KEY = "gram_size";
+ public static final String NGRAM_BF_SIZE_KEY = "bf_size";
Review Comment:
I think it's by design. The idea mainly from clickhouse, which I think
design for throughput. We could switch to set FPR instead of the bf size, but
which need keep the distinct values during write, that's cost too much.
BTW, if use the FPR, the end user may not clear about the cardinality of the
values, how about lead to big bloom filter size? which may hurt the performance
rather than improve.
So overall, I think it's better to keep the design.
--
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]
