Postgres table name limit on preaggregation tables

[lvauvillier]

Describe the bug

Postgres has a limitation of 64 chars for table names.
Generated pre-aggregation table names can reach this limit:

        Schema        |                              Name                               | Type  |  Owner   
----------------------+-----------------------------------------------------------------+-------+----------
 stb_pre_aggregations | impressions_advertiser_count_by_day20200601_hnkacipl_vhthxxhc_1 | table | postgres
 stb_pre_aggregations | impressions_advertiser_count_by_day20200701_wzp3uyld_qnhk3tsr_1 | table | postgres
 stb_pre_aggregations | impressions_publisher_count_by_day20200601_kmpd5gs3_x0zn2l0v_15 | table | postgres
 stb_pre_aggregations | impressions_publisher_count_by_day20200701_1zpylxe_uma3nxf1_159 | table | postgres
 stb_pre_aggregations | requests_fill_rate_by_day20200601_5ypc2raf_gevo3jro_15954312366 | table | postgres
 stb_pre_aggregations | requests_fill_rate_by_day20200701_jsp0kf3_d4klpmk1_159572190821 | table | postgres

In this case, timestamps are not valids and runs issue in preaggregation computations.

Maybe generate a fixed hash for the (cube, preAggregationName) part to avoid reaching this limit?

Version:
[0.19.39]

[vasilev-alex]

hey @lvauvillier! Thanks for posting this one. There's a similar issue #86.

As a workaround you can use sqlAlias for member names https://cube.dev/docs/cube#parameters-sql-alias