Uploaded image for project: 'MariaDB ColumnStore'
  1. MariaDB ColumnStore
  2. MCOL-4529

Design the proper way of doing extent elimination on varchars with utf8 or utf8mb4

    XMLWordPrintable

Details

    • Task
    • Status: Closed (View Workflow)
    • Critical
    • Resolution: Done
    • None
    • N/A
    • Documentation
    • None

    Description

      Overall - the top ticket (4522) is a massive regression.

      Starting in 5.5, the true length of varchar and char conforms to SQL standard, and incorporates byte counts for character sets such as utf8, and utf8mb4. Up to and including 5.4, it was wrong, and only reserved the bytes equal to the number of characters defined in the column DDL. The side effect of it is exemplified in the customer complaint: a varchar(5) under olf utf8 used to fit, and filters on it were subject to extent elimination. With 5.5, it no longer fits, which causes columnstore to essentially ignore an otherwise tight filter, and to proceed with the full scan.

      Attachments

        Issue Links

          Activity

            People

              sergey.zefirov Sergey Zefirov
              gdorman Gregory Dorman (Inactive)
              Votes:
              0 Vote for this issue
              Watchers:
              3 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved:

                Git Integration

                  Error rendering 'com.xiplink.jira.git.jira_git_plugin:git-issue-webpanel'. Please contact your Jira administrators.