resolve std name conflict

[ghost]

Due to forward declarations and the structure of std includes I had compiler warning for redeclaring some atomic structures.
Here is a possible solution using defines and an internal namespace in std.

[dvyukov]

{ on new line please

[dvyukov]

What compiler are you using? And what warnings does it produce?

[ghost]

I had some std header included (vector, string) and on Visual Studio 2015 it was complaining about the memory_order:
error C2874: using-declaration causes a multiple declaration of 'rl::memory_order'.
The include hierarchy was something like: vector included xmemory which included xatomic.

On GCC (5, the current mainstream gcc compiler on ubuntu) it had a similar problem with the atomic.
I have not checked that.

[dvyukov]

Does mutex lead to the same problems?
It's pretty common name, so redefining it is unfortunate...

[ghost]

It was a fast search/replace and did not considered any other name conflict.
I know it's not the best solution but had no better idea.
The best would be to pretend that atomic, thread, mutex, etc. has already been included, but I doubt if it has a solution independent of the implementation of libstd.

Note, I had the define problem with delete too.

class Foo {
  Foo(const Foo&) = delete;
}

[dvyukov]

It was a fast search/replace and did not considered any other name conflict.

So does mutex lead to the same problems or not?

Foo(const Foo&) = delete;

Just don't use this with Relacy.
Define-based instrumentation sucks, it all needs to be redone to use compiler instrumentation. But I don't have time for that.

[ghost]

mutex was ok, did not have the problem.
Unfortunatelly I forget to create a branch before push/PR. So I'll close the PR and create a new branch.
Are u interested in the patch ? Shall I resend the PR from the new branch?

[dvyukov]

Unfortunatelly I forget to create a branch before push/PR. So I'll close the PR and create a new branch.

git push -f?

Are u interested in the patch ?

Yes. It fixes build for you, right?
Without mutex/condvar changes it looks good to me.

Thanks