Sample CellML files for testing

[jonc125]

For #21. I think this is now a fairly complete set of tests for the CellML parsing side of things.

[tamuri]

Looks good. Well-formed, MathML valid and transpiler works, RDF valid. We can check the various connections/annotations as we develop. Out of interest, the CellML XML schema has for connection type:

<xsd:element name="connection" type="connection_type" />

<xsd:complexType name="connection_type">
    <xsd:choice  maxOccurs="unbounded" >
      <xsd:element ref="map_components" />
      <xsd:element ref="map_variables" />
    </xsd:choice>
  </xsd:complexType>

Therefore, this test model (and the weblab cellml examples I looked at) doesn't validate with schema.

[jonc125]

I'm aware that the XML Schema for CellML 1 doesn't capture all restrictions, but I thought it would at least validate things that were valid. Though I've not tried it in years. I checked the files with pycml and it says they're OK, which I trust much more :)

Though I don't see why the snippet you showed would fail to validate - doesn't the maxOccurs on the choice let you have more than 1 of each? So it's really too lenient (you should have at most one map_components).

[tamuri]

<xsd:choice> means only one of the elements, but that one element can occur multiple times with maxOccurs="unbounded". The CellML DTD also defines the same choice but has the occurence as you describe: <!ELEMENT connection (map_components | map_variables+)>

[jonc125]

Not if the maxOccurs is on the choice - that means that the choice itself can be made many times, and each time it's free to choose differently.

[jonc125]

If the maxOccurs was on each child element declaration, then it'd do what you're describing.

[tamuri]

Ah, yes, that's right.

[tamuri]

I checked the files with pycml and it says they're OK, which I trust much more :)

Thanks & confirmed, validates with the Relax NG schema from https://chaste.comlab.ox.ac.uk/cellml/