Uploaded image for project: 'Lucene - Core'
  1. Lucene - Core
  2. LUCENE-3738

Be consistent about negative vInt/vLong

Details

    • Bug
    • Status: Closed
    • Blocker
    • Resolution: Fixed
    • None
    • 3.6, 4.0-ALPHA
    • None
    • None
    • New

    Description

      Today, write/readVInt "allows" a negative int, in that it will encode and decode correctly, just horribly inefficiently (5 bytes).

      However, read/writeVLong fails (trips an assert).

      I'd prefer that both vInt/vLong trip an assert if you ever try to write a negative number... it's badly trappy today. But, unfortunately, we sometimes rely on this... had we had this assert in 'since the beginning' we could have avoided that.

      So, if we can't add that assert in today, I think we should at least fix readVLong to handle negative longs... but then you quietly spend 9 bytes (even more trappy!).

      Attachments

        1. ByteArrayDataInput.java.patch
          4 kB
          Uwe Schindler
        2. LUCENE-3738.patch
          14 kB
          Uwe Schindler
        3. LUCENE-3738.patch
          11 kB
          Uwe Schindler
        4. LUCENE-3738.patch
          1 kB
          Michael McCandless
        5. LUCENE-3738.patch
          1 kB
          Uwe Schindler
        6. LUCENE-3738.patch
          0.6 kB
          Uwe Schindler
        7. LUCENE-3738-improvement.patch
          7 kB
          Uwe Schindler

        Issue Links

          Activity

            People

              uschindler Uwe Schindler
              mikemccand Michael McCandless
              Votes:
              0 Vote for this issue
              Watchers:
              1 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: