cube-js/cube · error · UserError

Not implemented yet

Error message

Not implemented yet

What it means

CrateDB adapter inherits Postgres but does not implement HLL (HyperLogLog) initialization SQL, so hllInit unconditionally throws 'Not implemented yet'. Approximate distinct counting that requires HLL init on the source dialect cannot run on CrateDB.

Source

Thrown at packages/cubejs-schema-compiler/src/adapter/CrateQuery.ts:6

import { PostgresQuery } from './PostgresQuery';
import { UserError } from '../compiler/UserError';

export class CrateQuery extends PostgresQuery {
  public hllInit(_sql): string {
    throw new UserError('Not implemented yet');
  }

  public hllMerge(_sql): string {
    throw new UserError('Not implemented yet');
  }

  public countDistinctApprox(sql: string): string {
    return `hyperloglog_distinct(${sql})`;
  }

  public sqlTemplates() {
    const templates = super.sqlTemplates();
    delete templates.functions.WIDTH_BUCKET;
    return templates;
  }
}

View on GitHub (pinned to 7d981676b3)

Solutions

  1. Use exact distinct count instead of the approximate variant (plain `type: 'countDistinct'`).
  2. Use CrateDB's native `hyperloglog_distinct` path via countDistinctApprox which is implemented (hllMerge/hllInit are not).
  3. Contribute an implementation of hllInit in a custom query subclass extending CrateQuery.

Example fix

// before
measures: { users: { type: 'countDistinctApprox', sql: `${CUBE}.userId` } }
// after
measures: { users: { type: 'countDistinct', sql: `${CUBE}.userId` } }
Defensive patterns

Strategy: fallback

Validate before calling

if (dbType === 'cratedb' && measure.type === 'countDistinctApprox') {
  console.warn('HLL init not implemented for CrateDB; use countDistinct');
}

Try / catch

try { await cube.load(q) } catch (e) { if (/Not implemented yet/.test(e.message)) { return cube.load({ ...q, measures: exactMeasures }); } throw e; }

Prevention

When it happens

Trigger: Calling countDistinctApprox-adjacent SQL generation that routes to hllInit on a CrateDB datasource — typically when an approximate distinct measure (approximateUnique style) triggers HLL aggregation in a sub-query.

Common situations: Using approximate distinct count measures in a project pointed at CrateDB; porting a Postgres schema using HLL extensions to CrateDB.

Related errors

AI-assisted analysis of cube-js/cube@7d981676b3 (2026-09-02). Data as JSON: /api/errors/730f618c93997bfb. Report an issue: GitHub.