Den 03-09-2011 22:19, Jonathan M Davis skrev:
On Saturday, September 03, 2011 22:13:44 jdrewsen wrote:
Den 01-09-2011 18:52, Timon Gehr skrev:
On 09/01/2011 06:39 PM, jdrewsen wrote:
Den 01-09-2011 05:19, David Nadlinger skrev:
On 8/25/11 12:05 PM, Jonas Drewsen wrote:
Actually would adding opCast!string() and opCast!(ubyte[])() to
AsyncResult and Result allow for this?:
string content = Http.get("http://mysite.com/form.cgi";);
or
ubyte[] content = Http.get("http://mysite.com/form.cgi";);
It wouldn't, opCast is only taken into account for explicit casts
(if(something) is considered an explicit cast to bool). For this to
work, you would have to use alias this, but currently you could only
provide one of the two conversions.
David
Ok. You can do it in C++. I wonder what this has not been allowed in
D?
/Jonas
Because in D there is/should be multiple alias this, which is a better
concept to solve the same set of problems.
Will multiple alias this is currently not supported and AFAIK not
planned. Furthermore I don't how doing alias this on both string and
ubyte[] is possible since they share a lot of interface. For example: If
the class Result has alias this for ubyte[] and for string then what
does this mean?
Result res = .....;
writeln("The length is ", res.length);
Therefore I still think implicit opCast would be nice. Maybe the opCast
could have an @implicit attribute or something to make i safer.
TDPL explicitly states that D supports multiple alias this. It just hasn't
been implemented in dmd yet. It is certainly true though that there may be
issues with having multiple alias this of types which are too similar, and
those issues have not necessarily been ironed out yet.
Ok. Must have got it wrong then.
opImplicitCast has been requested on a number of occasions, but there are not
currently any plans for such a feature. alias this is intended to fulfill that
role, and I don't see how opImplicitCast would really be any different from
alias this, since you can give a function to alias this already (and then the
type implicitly converts to the function's return type).
Didn't know the return type trick. Nice!
/Jonas
