Derby
  1. Derby
  2. DERBY-2892 Closing a resultset after retrieving a large > 32665 bytes value with Network Server does not release locks
  3. DERBY-2941

With 10.2, Closing a resultset after retrieving a large > 32665 bytes value with Network Server does not release locks

    Details

    • Urgency:
      Urgent

      Description

      Making a subtask for the 10.2 fix for this issue as the fix for 10.3 will be different.

      1. DERBY-2941_diff.txt
        14 kB
        Kathey Marsden

        Activity

        Hide
        Kathey Marsden added a comment -

        Closing this issue for the fix with 10.2 and lower clients. The problem still exists with 10.3 clients which use the lob locator code path. DERBY-2892 remains open for that issue.

        Show
        Kathey Marsden added a comment - Closing this issue for the fix with 10.2 and lower clients. The problem still exists with 10.3 clients which use the lob locator code path. DERBY-2892 remains open for that issue.
        Hide
        Kathey Marsden added a comment -

        I checked the fix into the 10.3 branch as well, so hopefully the fix version is ok. It only fixes 10.2 client with 10.3. It does not fix 2892 10.3 client with 10.3.

        Show
        Kathey Marsden added a comment - I checked the fix into the 10.3 branch as well, so hopefully the fix version is ok. It only fixes 10.2 client with 10.3. It does not fix 2892 10.3 client with 10.3.
        Hide
        Myrna van Lunteren added a comment - - edited

        I know you checked in the test to 10.3 branch, but I'm doubting that this issue should be marked fixed for 10.3.1.2?

        Show
        Myrna van Lunteren added a comment - - edited I know you checked in the test to 10.3 branch, but I'm doubting that this issue should be marked fixed for 10.3.1.2?
        Hide
        Kathey Marsden added a comment -

        Fixed in 10.2 and ported to trunk and 10.3 and verified with 10.2 client. DERBY-2892 still needs to be fixed for server working with 10.3+ client

        Kathey

        Show
        Kathey Marsden added a comment - Fixed in 10.2 and ported to trunk and 10.3 and verified with 10.2 client. DERBY-2892 still needs to be fixed for server working with 10.3+ client Kathey
        Hide
        Tomohito Nakayama added a comment -

        I remember that I had trouble to retrieve length of stored data in DERBY-326.
        http://mail-archives.apache.org/mod_mbox/db-derby-dev/200512.mbox/%3C43A975F5.3090904@basil.ocn.ne.jp%3E

        At that point, it seems that length information might be missing in some cases....

        My understanding is that the engine was improved and
        it became always possible to retrieve length of stored data via DataValueDescriptor#getLength().

        Show
        Tomohito Nakayama added a comment - I remember that I had trouble to retrieve length of stored data in DERBY-326 . http://mail-archives.apache.org/mod_mbox/db-derby-dev/200512.mbox/%3C43A975F5.3090904@basil.ocn.ne.jp%3E At that point, it seems that length information might be missing in some cases.... My understanding is that the engine was improved and it became always possible to retrieve length of stored data via DataValueDescriptor#getLength().
        Hide
        Øystein Grøvlen added a comment -

        I have looked at the patch, and I think the patch looks very good.
        I only have some minor comments/questions:

        • Did you intend to leave the changes to DDMWriter in there?
        • You have added a comment that says "Stream is initialized on the
          first read". It is not quite clear to me what stream you are
          referring to. It looks to me that the caller has to initialize the
          stream (call initInputStream) BEFORE it is to be read.
        • There seem to be some indentation differences in the
          EXTDTAInputStream constructor.
        • I do not understand the javadoc for EXTDTAInputStream.isNUll. What
          do you mean by "determined in the constructor"? It seems like the
          only thing the constructor does is to register the parameter
          values.

        +1 to commit, but would prefer that the comments discussed about is made clearer.

        Show
        Øystein Grøvlen added a comment - I have looked at the patch, and I think the patch looks very good. I only have some minor comments/questions: Did you intend to leave the changes to DDMWriter in there? You have added a comment that says "Stream is initialized on the first read". It is not quite clear to me what stream you are referring to. It looks to me that the caller has to initialize the stream (call initInputStream) BEFORE it is to be read. There seem to be some indentation differences in the EXTDTAInputStream constructor. I do not understand the javadoc for EXTDTAInputStream.isNUll. What do you mean by "determined in the constructor"? It seems like the only thing the constructor does is to register the parameter values. +1 to commit, but would prefer that the comments discussed about is made clearer.
        Hide
        Kathey Marsden added a comment -

        Attaching a patch for this issue. Patch only differs from patch attached to DERBY-2892 in that it has some corrections to javadoc.

        -Changes EXTDTAInputStream to use ResultSet.getCharacterStream, ResultSet.getBinaryStream instead of getClob(), getBlob..

        • Alters EngineResult for getLenth(int columnIndex) and isNull(int columnIndex) methods

        -Enables LargeDataLocksTest

        derbyall passed with only known failures.

        I would most appreciate review. If I hear no objections I will commit tomorrow.

        Show
        Kathey Marsden added a comment - Attaching a patch for this issue. Patch only differs from patch attached to DERBY-2892 in that it has some corrections to javadoc. -Changes EXTDTAInputStream to use ResultSet.getCharacterStream, ResultSet.getBinaryStream instead of getClob(), getBlob.. Alters EngineResult for getLenth(int columnIndex) and isNull(int columnIndex) methods -Enables LargeDataLocksTest derbyall passed with only known failures. I would most appreciate review. If I hear no objections I will commit tomorrow.

          People

          • Assignee:
            Kathey Marsden
            Reporter:
            Kathey Marsden
          • Votes:
            0 Vote for this issue
            Watchers:
            0 Start watching this issue

            Dates

            • Created:
              Updated:
              Resolved:

              Development