Support Array(LowCardinality(String)) for Bloom Filter

[volfco]

Use case
To save space I have a table with a large Array column (because I can't dynamically create columns when using the Kafka table engine) with a type of LowCardinality(String).

Altering this table to use a bloom filter index results in the following error:
Unexpected type Array(LowCardinality(String)) of bloom filter index..

Describe the solution you'd like
Support for bloom filter indexes on columns with the Array(LowCardinality(String)) type

[makeavish]

Is it going to be picked anytime soon?

[darkleaf]

It works on 22.3.

CREATE TABLE x
(
    `foo` Array(LowCardinality(String)),
    INDEX idx foo TYPE bloom_filter GRANULARITY 1
)
ENGINE = MergeTree
PRIMARY KEY tuple()

Query id: 50d364b8-77c8-4cd5-afa0-5ff8933d3932

Ok.

0 rows in set. Elapsed: 0.063 sec. 

insert into x values (['a', 'b']);

INSERT INTO x FORMAT Values

Query id: d5b3cbbd-c91e-487b-bd25-20cd9ab37289

Ok.

1 rows in set. Elapsed: 0.087 sec. 

SELECT *
FROM x

Query id: 0b054751-289f-4163-8c0a-46dabb1a0e15

┌─foo───────┐
│ ['a','b'] │
└───────────┘

1 rows in set. Elapsed: 0.050 sec. 

SELECT *
FROM system.data_skipping_indices
WHERE table = 'x'

Query id: 6fe12741-594f-4bd8-8661-a8b053aad869

┌─database───┬─table─┬─name─┬─type─────────┬─expr─┬─granularity─┬─data_compressed_bytes─┬─data_uncompressed_bytes─┬─marks─┐
│ gmonit-dev │ x     │ idx  │ bloom_filter │ foo  │           1 │                    28 │                       2 │    24 │
└────────────┴───────┴──────┴──────────────┴──────┴─────────────┴───────────────────────┴─────────────────────────┴───────┘