Improve definition of XExpr type

[joergen7]

The type XExpr has been defined as follows:

(define-type XExpr
  (U XML-Misc Cdata Positive-Index
     Number String Symbol
     (Pair Symbol (Pair (Listof XExpr-Attribute) (Listof XExpr)))
     (Pair Symbol (Listof XExpr))))

The above definition includes the superfluous type Number but is missing the type Pcdata.

The type Number is superfluous because so-called valid characters (valid-char?) are already covered by Positive-Index. Pcdata structs are covered by xexpr?, but are missing from the above type definition.

[github-actions]

Resyntax analyzed 0 files in this pull request and found no issues.

[samth]

I worry a little about restricting the space of numbers like this. The change means that (- 2 1) no longer typechecks as XExpr.

[joergen7]

I see your point. I had assumed this restriction to be a feature, since now you have to make an argument for the subtraction result to be positive for XExpr to type: (assert (- 2 1) positive?).

Which is preferable?

[joergen7]

The pr now introduces only Pcdata, thus, strictly widening the definition of XExpr.

[github-actions]

Resyntax analyzed 0 files in this pull request and found no issues.

[github-actions]

Resyntax analyzed 0 files in this pull request and found no issues.

[samth]

Unfortunately this goal is not possible. Consider 55926. It is a Positive-Index but not a valid-char?.

[joergen7]

I understand. Thanks, @samth for your patience. The current change strictly widens the definition of XExpr.

[github-actions]

Resyntax analyzed 0 files in this pull request and found no issues.