Re: [pygccxml-development] Two patches proposal

[Roman Y. <rom...@gm...>]

On Wed, Dec 23, 2009 at 4:04 PM, Berserker <ber...@ho...> wrote:
> Yes, I'm actually using that solution. I'm sad about "user expectations" but
> what
> about this case then (without the patch):
>
> foo = Foo()
> b1 = foo.Bar
> b1.x = 2
> b2 = foo.Bar
> b2.x = ?
>
> I think that this is another case where we can discuss about "user
> expectations", what about?

Look, I don't want to argue - my experience is different from yours. I
believe, in this case Py++ provides a good way to change the default
behavior. The code you are using is in the "public interface" and I am
not going to break it without very, very good reason.

> I'm not a "standard expert" too, but I can confirm that your code does
> compile
> on MSVC (tested on VC2005 and VC2008).

Thanks.

I read MSDN (http://msdn.microsoft.com/en-us/library/s2ff0fz8%28VS.100%29.aspx)
and it defines "conversion constructor" as "...A constructor that can
be called with a single argument is used for conversions from the type
of the argument to the class type. ..."

I changed and committed the code:
http://pygccxml.svn.sourceforge.net/viewvc/pygccxml?view=rev&revision=1782


> Actually I "tweaked" the method in this way (probably it could be written
> better...):

Yes :-) take a look on the difference
>
> decl_wrappers.calldef_wrapper.constructor_t.does_define_implicit_conversion
> = override_does_define_implicit_conversion
>
> It works but as you probably know I don't like this solution :)

In this case I completely agree with you and hoe you will like SVN
version better.

> Thanks for your help as usual ;)

You are welcome.

-- 
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/