<degree> can be child of other stuff?

[kerimoyle]

See https://github.com/cellml/libcellml/pull/631/files#r448700802

[agarny]

Nice digging job! :)

[nickerso]

The spec is never wrong! In CellML, the degree element is only allowed as the child of root and diff expressions - there is no implication that the sentence means a child in terms of the XML serialisation.

[agarny]

Hmm... I still find it confusing. I would personally rephrase it to read "degree MUST be used in the context of a root or diff", or something like that, but spec writing clearly eludes me, so...

[kerimoyle]

If the spec is correct (however it's phrased) then the check has to happen in the Analyser/Validator instead?

[agarny]

The validator validates all the maths using the W3C MathML DTD, so we are fine.

[kerimoyle]

It doesn't check this CellML-specific restriction though ... which was the point of the discussion.

[nickerso]

Does it check that only the permitted MathML elements are allowed? In which case it would be checking this restriction by implication...

[kerimoyle]

It doesn't check where it happens, only that the tag is one of the CellML set (see below). The Analyser is prob the best place to check it as it's the only place where the mathml is actually interpreted into anything?

static const std::vector<std::string> supportedMathMLElements = {
    "ci", "cn", "sep", "apply", "piecewise", "piece", "otherwise", "eq", "neq", "gt", "lt", "geq", "leq", "and", "or",
    "xor", "not", "plus", "minus", "times", "divide", "power", "root", "abs", "exp", "ln", "log", "floor",
    "ceiling", "min", "max", "rem", "diff", "bvar", "logbase", "degree", "sin", "cos", "tan", "sec", "csc",
    "cot", "sinh", "cosh", "tanh", "sech", "csch", "coth", "arcsin", "arccos", "arctan", "arcsec", "arccsc",
    "arccot", "arcsinh", "arccosh", "arctanh", "arcsech", "arccsch", "arccoth", "pi", "exponentiale",
    "notanumber", "infinity", "true", "false"};

The DTD error is way too general for our specific case:

W3C MathML DTD error: Element apply content does not follow the DTD, expecting (csymbol | ci | cn | apply | reln | lambda | condition | declare | sep | semantics | annotation | annotation-xml | integers | reals | rationals | naturalnumbers | complexes | primes | exponentiale | imaginaryi | notanumber | true | false | emptyset | pi | eulergamma | infinity | interval | list | matrix | matrixrow | set | vector | piecewise | lowlimit | uplimit | bvar | degree | logbase | momentabout | domainofapplication | inverse | ident | domain | codomain | image | abs | conjugate | exp | factorial | arg | real | imaginary | floor | ceiling | not | ln | sin | cos | tan | sec | csc | cot | sinh | cosh | tanh | sech | csch | coth | arcsin | arccos | arctan | arccosh | arccot | arccoth | arccsc | arccsch | arcsec | arcsech | arcsinh | arctanh | determinant | transpose | card | quotient | divide | power | rem | implies | vectorproduct | scalarproduct | outerproduct | setdiff | fn | compose | plus | times | max | min | gcd | lcm | and | or | xor | union | intersect | cartesianproduct | mean | sdev | variance | median | mode | selector | root | minus | log | int | diff | partialdiff | divergence | grad | curl | laplacian | sum | product | limit | moment | exists | forall | neq | factorof | in | notin | notsubset | notprsubset | tendsto | eq | leq | lt | geq | gt | equivalent | approx | subset | prsubset | mi | mn | mo | mtext | ms | mspace | mrow | mfrac | msqrt | mroot | menclose | mstyle | merror | mpadded | mphantom | mfenced | msub | msup | msubsup | munder | mover | munderover | mmultiscripts | mtable | mtr | mlabeledtr | mtd | maligngroup | malignmark | maction)*, got (CDATA bvar ).

[nickerso]

if the DTD + restricted set of elements isn't enough, then I'd suggest this is something the validator should be checking in future as we look to make the AST more widely accessible. The analyser shouldn't be validating - that is why it will only work on valid models, right?

[agarny]

if the DTD + restricted set of elements isn't enough, then I'd suggest this is something the validator should be checking in future as we look to make the AST more widely accessible. The analyser shouldn't be validating - that is why it will only work on valid models, right?

I would indeed expect the Validator to do that kind of check, if deemed necessary.

[kerimoyle]

Closing in favour of libcellml-based discussion here

[kerimoyle]

The MathML2 manual says that this is valid:

<apply>
  <diff/>
  <bvar>
    <ci> x </ci>
    <degree><cn> 2 </cn></degree>
  </bvar>
  <apply><fn><ci>f</ci></fn>
    <ci> x </ci>
  </apply>
</apply>

In other words, <degree> is a direct child of <bvar> rather than <diff>. Is this ok in CellML? I'm struggling to see how it's allowed as a child of diff and not a child of bvar too ... ?