OPENXML to get content by using element and attribute name

In this SQL Server OPENXML example:

declare @xmldata xml    

set @xmldata = 
'

    
        Task 1
        2020-11-14
    
    
        2020-11-15
        Task 2
    
'

DECLARE @hDoc int
EXEC sp_xml_preparedocument @hDoc OUTPUT, @xmldata

SELECT *
FROM OPENXML(@hDoc, '//wddxPacket/data',2)
        WITH(
            task          varchar(200) 'var[1]/string',
            startdate     varchar(200) 'var[2]/string'
    )

EXEC sp_xml_removedocument @hDoc

This is the current output:

task         startdate
------       ----------
Task 1       2020-11-14
2020-11-15   Task 2

But I really want this output:

task         startdate
------       ----------
Task 1       2020-11-14
Task 2       2020-11-15

To identify the content, I cannot use "var" alone, because all elements of interest are named "var". I cannot use position [1] or [2], because the position numbers are inconsistent in the XML, resulting in the current mixed-up output. How do I identify the elements based on "var" and the "dimension" attribute names of "task" or "startdate"?

You need to use a different XPath expression to match the attribute values, eg var[@dimension="task"]/string.

Like this:

declare @xmldata xml    

set @xmldata = 
'

    
        Task 1
        2020-11-14
    
    
        2020-11-15
        Task 2
    
'

DECLARE @hDoc int
EXEC sp_xml_preparedocument @hDoc OUTPUT, @xmldata

SELECT *
FROM OPENXML(@hDoc, '/wddxPacket/data',2)
        WITH(
            task          varchar(200) 'var[@dimension="task"]/string',
            startdate     varchar(200) 'var[@dimension="startdate"]/string'
    )

EXEC sp_xml_removedocument @hDoc

outputs

task    startdate
------- ------------
Task 1  2020-11-14
Task 2  2020-11-15

StackExchange Database Administrators Q#279695, answer score: 8