Re: [Nmock-general] FW: NMock, MSTest and Interface Implementation

[Steve M. <Ste...@ty...>]

The IL generation is pretty straightforward wrapper of the interfaces.
The reason it is required is that certain operations go through the
proxy and ask for the type of the underlying object.  The proxy is
responding with 'MockObject' in the original code.  The code is actually
pretty much borrowed straight over from the multi-interface factory,
because at the time I didn't quite understand everything that was going
on in the library.

I currently believe that with a little extra work the Multi-interface
factory and the new Mock object patch could be combined into the same
facility, or at least worked a bit to share a bit more code.

-----Original Message-----
From: Mike Capp [mailto:mik...@gm...]=20
Sent: Monday, August 06, 2007 12:46 PM
To: Steve Mitcham; nmo...@li...
Subject: Re: FW: [Nmock-general] NMock, MSTest and Interface
Implementation

(Wasn't subscribed to nmock-general, and the web archive seems to be
very tardy.)

> -----Original Message-----
> [mailto:nmo...@li...] On Behalf Of
Steve
> Freeman
> Sent: Tuesday, July 24, 2007 9:42 AM
>
> At this point, you guys /are/ the contributors. Mike, what do you
> think about the patch?

I'm fine with the patch in that it fixes a problem passes its own
tests and didn't break any of ours. My reservations were just that:

1) IIRC it's going back to runtime IL generation, and I don't have
enough of a mental model of how that works to predict with confidence
what will or won't or might be affected.

2) Given that this was pretty much how NMock 1 worked, I assumed that
Nat had reasons (aside from neophilia) for going the transparent proxy
route with NMock 2. Without knowing what those reasons were, it's hard
to weigh them.

Mike