Re: non-prototype function definitions

[Roland McGrath <roland at hack dot frob dot com>]

So it was you who broke the enforcement of the second property that we
used to get.  Would you like now to be part of the solution to the
problem you created?  

I'm not aware of any violations of the two rules I stated in the libc
ABI.  If there are any that have crept in, we should deprecate the
broken function signatures and replace them with new ones.
Regardless, we should find a way to ensure that no such errors are
introduced.

I do indeed care that we preserve that second property.  I wasn't
aware until today that you'd broken the useful feature of the compiler
that we used to enforce it in the past.  We need a new way to ensure
that no such signatures creep into the ABI in the future.


Thanks,
Roland