On 2011-06-19 13:28, Stewart Gordon wrote:
> On 18/06/2011 11:52, Mafi wrote:
> > Am 18.06.2011 10:36, schrieb Jonathan M Davis:
> <snip>
> 
> > If anyone ever writes a function which needs immutable chars, toStringz
> > just wroks for it.
> 
> <snip>
> 
> You mean a C function that keeps the pointer for later use, and relies on
> you not to change it?

That doesn't require immutable at all. And returning const(char)* would work 
just as well. You can't alter it in either case. The only way that it would be 
alterable would be if the original string which was passed to toStringz wasn't 
immutable (e.g. char[]), and the overload of toStringz which took it didn't 
make a new array. But the overload of toStringz which takes a char[] _does_ 
return a new array. So, there is _no_ difference between returning 
const(char)* and immutable(char)* unless you have a D function which takes 
immutable(char)* that you're trying to pass it to.

Of course, in the case that you're describing, there's always the fun bug that 
the GC doesn't know that it needs to hang onto the immutable(char)* which was 
returned unless you hold a reference to it somewhere, in which case it could 
be collected and screw over the C function which held onto it. Now that I 
think about it, it might actually be worth adding a note to the documentation 
about that.

- Jonathan M Davis

Reply via email to