>>33
Glossing over the fact that you can't really check types very well in the first place, and for the third time: everything that isn't a type can't be checked. You don't need an example, just take some code and at each part ask yourself "is this a type?" - if the answer is "no," then it is something that can't be checked semantically. Now you can make up all the examples you'd like. There are infinitely many of them and they exist in all C++ code.
You can't even conceive of a language whose type system can be used to check semantics, can you? For the third time: learn another language already.
>>35
This started when some jag said you could make C++ as safe as anything else. It's not about the need for safety, but the availability of it. Whether you (or I, or anyone) like it or want it or think anyone should have it is irrelevant at this point.