[expr.type.conv]/2: Priority of first "if", and binding of "otherwise"

[hubert-reinterpretcast]

In subclause 8.2.3 [expr.type.conv] paragraph 2 of the 2017 DIS, the structure of:

If the initializer is a parenthesized single expression, the type conversion expression is equivalent (in definedness, and if defined in meaning) to the corresponding cast expression ([expr.cast]).
If the type is cv void and the initializer is (), [ ... ].
Otherwise, [an] object is direct-initialized (11.6) with the initializer.

Presents a interpretation where the first sentence is elaborated upon by the latter two as opposed to covering disjoint cases.

Given:

int *f(void *p) {
  typedef int *IntPtr;
  return IntPtr(p);
}

The C-style cast interpretation is valid; however, the direct-initialization is not.
Thus the first sentence conflicts with the latter two under said interpretation.

Furthermore, the parenthetical in the first sentence is meant to be parsed as:

equivalent (in definedness; and if defined, in meaning) to the corresponding cast expression

However, it may be taken as:

equivalent in definedness, should the corresponding cast expression be defined

Suggestion:
Move the latter two sentences under an "Otherwise" bound to the first, with appropriate indentation.
Indent the "then" part of the first as well.

Suggestion:
Improve punctuation within the parenthetical in the first sentence.

[jensmaurer]

I think starting the second sentence with "Otherwise" already helps: It presents a clear "If ... Otherwise, if ... Otherwise, ..." ladder. Also, simply striking the rather strange parenthetical remark seems not to lose anything and fixes the hostile interpretation issue.