+X/Open curses documented these functions to pass an array of \fBwchar_t\fR
+in 1997, but that was an error because of this part of the description:
+.RS
+.PP
+The effect of \fIget_wstr()\fP is as though a series of calls to
+\fIget_wch()\fP were made, until a newline character, end-of-line character, or
+end-of-file character is processed.
+.RE
+.PP
+The latter function \fIget_wch()\fP can return a negative value,
+while \fBwchar_t\fP is a unsigned type.
+All of the vendors implement this using \fBwint_t\fR, following the standard.
+.PP
+X/Open Curses issue 7 is unclear regarding whether
+the terminating \fInull \fP\fBwchar_t\fP
+value is counted in the length parameter \fIn\fP.
+X/Open Curses issue 7 revised the corresponding description
+of \fBwgetnstr\fP to address this issue.
+The unrevised description of \fBwget_nwstr\fP can be interpreted either way.
+This implementation counts the terminator in the length.
+.PP
+X/Open Curses does not specify what happens if the length \fIn\fP is negative.
+.bP
+For analogy with \fBwgetnstr\fP,
+ncurses 6.2 uses a limit (based on \fBLINE_MAX\fP).
+.bP
+Some other implementations (such as Solaris xcurses) do the same,
+while others (PDCurses) do not allow this.
+.bP
+NetBSD 7 curses imitates ncurses 6.1 in this regard,
+treating a \fB\-1\fP as an indefinite number of characters.