All tests pass for me on Visual Studio 2010 Premium. I don't have objections to anything in the patch. The only thing that might be controversial, from what I can see, is that since we're compiling using VC++ under Windows, we have to include many more pointer casts. In particular, we have to cast the result of all of our malloc-like calls, since C++ doesn't automatically cast from void *. In C, the best practice is to not include a cast, so that the compiler can yell at you if you forgot to include the header file that defines malloc. I think that since we're using avro_malloc and friends, rather than the system malloc, this is not an issue that we have to worry about.
So, short version, +1. If I don't hear any objections from Bruce or others by Friday, I'll commit this is SVN.