#47 JQTI: Further issues with pre-escapes source XML

[Dave McKain]

JQTI effectively pre-escapes source XML so that entity and character references end up in the data model represented by their raw mark-up. I noted an issue with CDATA sections (privately) to Jonathon Hare earlier in the year, and have now experienced another couple of bugs as a result of this behaviour.

The first problem can be seen in bug1.xml (uploaded). Here, I've declared the correct response in the XML to be 2, which is of course a long-winded way of authoring the character '2' inside XML. The same character reference is used in the mapping to determine the scope. However, if the candidate enters a number 2 in the textEntryInteraction, this is deemed to be incorrect as the result of comparing the strings '2' and '2' is false, even though they are equal as far as XML is concerned. (This example is arguably rather contrived!)

A similar example which identifies a second issue is bug2.zip (also uploaded). This test contains basically the same item as above, but prompts the candidate to enter a '<' character, which has been entered in the XML as &lt;. If the candidate enters a literal '<' in the box (as you would hope for), the response is also judged to be incorrect in the same way as with the 1st example. However, additionally, if you then request an <assessmentTestResult>, the XML generated from this is not well-formed, as it contains a literal '<' sign:

<candidateResponse>
<value><</value>
</candidateResponse>

(To see this, upload the test into QTIEngine, play it, enter '<', then look at the report.)

[Dave McKain]

See main description

[Dave McKain]

See main description