MEMPTR memory leak - false positives

[Ton-Pe]

Case 1:

Following code gives a false positive. Size of memptr is set inside the block and then released outside that block.

DEFINE VARIABLE mReceiveData AS MEMPTR   NO-UNDO.

IF TRUE THEN DO:
    SET-SIZE(mReceiveData) = 100.
END.

SET-SIZE(mReceiveData) = 0.

If release is done inside the same block or moved to finally-block, then issue is not raised.

Case 2:

Following code where copy-lob is done inside a nested block, gives false positive.

DEFINE VARIABLE mReceiveData AS MEMPTR   NO-UNDO.
DEFINE VARIABLE mSendData    AS MEMPTR   NO-UNDO.

IF TRUE THEN DO: // First level
    DO: // second level
        COPY-LOB mReceiveData TO mSendData.
    END.
END.


FINALLY:
    SET-SIZE (mSendData) = 0.
END FINALLY.

For a comparison, this code, where there is only one block, issue is not raised:

DEFINE VARIABLE mReceiveData AS MEMPTR   NO-UNDO.
DEFINE VARIABLE mSendData    AS MEMPTR   NO-UNDO.

IF TRUE THEN DO:
        COPY-LOB mReceiveData TO mSendData.
END.

FINALLY:
    SET-SIZE (mSendData) = 0.
END FINALLY.

This is similar to issue #1121 in DynamicObjectLeak rule

[Philippe-La]

Case 1:

Following code gives a false positive. Size of memptr is set inside the block and then released outside that block.

DEFINE VARIABLE mReceiveData AS MEMPTR   NO-UNDO.

IF TRUE THEN DO:
    SET-SIZE(mReceiveData) = 100.
END.

SET-SIZE(mReceiveData) = 0.

If release is done inside the same block or moved to finally-block, then issue is not raised.

+1

[gquerret]

Fixed in develop branch, will be included in next release