[DRILL-2721] Identify, fix _existing_ INFO._SCHEMA columns in conflict with SQL spec. by 1.0

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.2.0
• Component/s: Metadata
• Assignee: Daniel Barclay
• Reporter: Daniel Barclay
• Created: 08/Apr/15 17:59
• Updated: 04/Jun/15 22:53
• Resolved: 04/Jun/15 22:53

Description

Much of the INFORMATION_SCHEMA.COLUMNS table doesn't follow the SQL specification.

Most cases are columns that simply aren't implemented yet, because they are of lower priority, probably because they won't be useful in Drill.

However, some columns that already exist conflict with what the SQL specification specifies for them. (For example, non-applicable precision values are represented as -1 rather than NULL.)

Those existing columns should probably be corrected before GA/version 1.0, so that users don't start depending on invalid values, which will make it harder to correct things later).

TBD: enumeration of specific problems/corrections
(rough notes of current awareness:

• some -1 should be NULL
• some -1/NULL for numeric precision should be 0
• floating-point numeric precision should be in bits, not digits
• it might be that type names should be canonical names, not syntactic-shortcut names (e.g., "VARCHAR" -> "CHARACTER_VARYING")
• character-type length and numeric precision might be wrongly combined into same column
  )

Issue Links

• is superceded by: DRILL-2763 [umbrella] Implement INFORMATION_SCHEMA.COLUMNS enough for relevant tools
• is superceded by: DRILL-3216 Fix existing(+) INFORMATION_SCHEMA.COLUMNS columns