Sergei Petrunia

... see the suggestion in : https://jira.mariadb.org/browse/MDEV-32403?focusedCommentId=284537&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-284537

> > I think it also makes sense to introduce `r_loops` in `subquery_materialization` (to be `materialization`) > > I don't understand what should be counted/tracked under this counter. Every lookup...

Also, note that for non-NULL-aware materialization, it reports `r_loops==r_index_lookup_loops`. "materialization": { "r_strategy": "index_lookup", "r_loops": 10, "r_index_lookup_loops": 10, "query_block": {

@Olernov , ok I think there are enough comments to get the next version of the patch.

> Do you mean for the `index_lookup` strategy? Yes, and it seems reasonable. But numbers differ for `partial match` strategies. Or you suggest removing redundant output for `index_lookup` strategy? Let's...

The result looks good. Please apply this code cleanup patch: https://gist.github.com/spetrunia/a09ad18f970d4ebe146e8cea5ed28bbd and update the test results.