[cfe-dev] Discussion: should we enforce access control in C++ attributes?

[Delesley Hutchins]

> You asked for opinions about the language design, and I offered mine.

I appreciate your input -- I got some good ideas for new features from
the discussion.  Thanks!

> Ultimately, I don't have any responsibility to your users, and you do.
> If you feel that it's just punishing your users to enable access control
> in thread-safety annotations, then go ahead and turn it off.

I do have a responsibility to my users, but I also have a
responsibility to the open source community.  I personally have mixed
feelings.  I want my users to be happy, but I also want thread-safety
attributes to be a well-designed language extension that follows the
rules of C++.  I figure cfe-dev is the right place to get opinions on
the latter; if the clang community thinks that keeping access control
turned on is the right idea, then I will keep it on, and tell my users
to use the workaround.

BTW, if it seems like I am arguing both sides of the case, I am; my
objective here is to foster discussion, and see what people think.
:-)

Thanks again,

  -DeLesley

-- 
DeLesley Hutchins | Software Engineer | delesley at google.com | 505-206-0315