Return the reason why the XML was not valid according to the XSD

[joaquinvanschoren]

No description provided.

[joaquinvanschoren]

Puzzlingly, the error returned is not always about the XSD at all, but is sometimes about the feature names. See, for instance: https://www.openml.org/api/v1/data/42075

[joaquinvanschoren]

To be more clear, this is what sometimes happens:

• A dataset has an OK XML description, but the ARFF file has 'forbidden' feature names.
• I thought that checking the feature names and the XML are two separate processes. However, what happens is that the error "illegal feature name" is returned when I check the XML file.
• Here is an example (check the error message): https://www.openml.org/api/v1/data/42075

What I think is happening is that:

• first the dataset file is checked and an error about the feature name is dropped in the libxml error buffer
• the XML is checked, and libxml_get_errors() notices that there is an error in the buffer, which is then returned
• the resulting error message says "the XML is invalid" together with the older error message about the feature names.

[TwsThomas]

Hi, what is the state of this Pull Request ?
Could it be merged ?