FWD » Core Development » Base Language

When READ-XML is used on a dynamic TEMP-TABLE before TEMP-TABLE-PREPARE, a NullPointerException is thrown.

ABL CodeABL Code

USING OpenEdge.Core.Assert FROM PROPATH.

CLASS tests.TestReadXmlTempTable:
    DEFINE PROTECTED VARIABLE ttData  AS HANDLE NO-UNDO.

    @Test.
    METHOD PUBLIC VOID TestReadXml():
        DEFINE VARIABLE res AS LOGICAL NO-UNDO.
        CREATE TEMP-TABLE ttData.
        res = ttData:READ-XML("FILE", "sample.xml", "EMPTY", ?, ?).
        Assert:IsTrue(res).
    END METHOD.
END CLASS.

Constantin Asofiei wrote:

Teodor, I don't think this approach is good - we know that READ-XML uses the buffer's (default or explicit) options, and not the temp-table.

What do you mean on this? I have analyzed the following cases:
- When BUFFER:READ-XML is called, the getXmlNodeName() is always not null, because of the getter implementation (also see #9763-40).
- When TEMP-TABLE:READ-XML is called and called before TEMP-TABLE-PREPARE, the override parameter is null and _isPrepared false. The TEMP-TABLE:XML-NODE-NAME is used because there is no default buffer.

I have also analyzed the changes from #9763 and I have noticed that the buffer parameter is only used for getXmlNodeName on importTable method and not anywhere else.
I think do you mean about the case when TEMP-TABLE was prepared and the default buffer XML-NODE-NAME was changed. I am looking on this right now.