W3C home > Mailing lists > Public > public-script-coord@w3.org > October to December 2012

[WebIDL] Typing of integer-token optional argument default values and constant values is unclear

From: Boris Zbarsky <bzbarsky@MIT.EDU>
Date: Thu, 01 Nov 2012 11:56:31 -0400
Message-ID: <50929BAF.6000109@mit.edu>
To: "public-script-coord@w3.org" <public-script-coord@w3.org>
The spec says in http://dev.w3.org/2006/webapi/WebIDL/#dfn-constant :

   The type of an integer token is the same as the type of the constant,
   dictionary member or optional argument it is being used as the value
   of.

   The value of the integer token MUST NOT lie outside the valid range
   of values for its type, as given in section 3.10 below.

and in http://dev.w3.org/2006/webapi/WebIDL/#idl-operations :

   When a boolean literal token (true or false), the null token, an
   integer token, a float token or one of the three special floating
   point literal values (Infinity, -Infinity or NaN) is used as the
   default value, it is interpreted in the same way as for a constant.

but on the face of it this is bogus, since this:

   void foo(optional DOMString something = 5);

should presumably not be valid, right?  And then this further raises the 
question of whether:

   void foo(optional float something = 5);

or

   const float myConst = 5;

are valid.  I think these _should_ be.  So I would propose that we 
explicitly limit the "same as the type" to "primitive types excluding 
boolean".  Furthermore in the float/double cases I think we should 
require that the integer be exactly representable.

-Boris
Received on Thursday, 1 November 2012 15:57:00 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 8 May 2013 19:30:07 UTC