On 05/05/2011 09:37 AM, Nikolay Sivov wrote:
On 5/5/2011 18:14, Adam Martinson wrote:
I think this is actually the most straightforward/readable way to handle it. If multi-dimensional SAFEARRAYs are actually supported, we will need separate cases for them.
It's just not the way I thought about it, that's all. Ok, let it be like that, but could you remove ERR() from it? I feel like it will be useless line that never works, on the other hand we should add dumping of array dimension in debugstr_variant() so it will be instantly visible in trace.
It's code that *shouldn't* ever execute AFAIK, that's why I used ERR().
And IMHO if an app is trying to do something we don't handle, and don't know how it should be handled, it should be a FIXME.
Well ok, I don't really expect it to come often.
I don't either, but it makes debugging a lot easier if it ever does, lol. I think it's just good practice for API functions to complain when they don't know how to handle their args.