Re: Comments in %union processed incorrectly

[Akim Demaille]

>>>>> "Paul" == Paul Eggert <address@hidden> writes:

Paul> If I understand you correctly, in C++ one cannot safely use
Paul> memcpy, alloca, malloc, or free, as all of these primitives
Paul> operate at a low level (merely copying the data bytes, as it
Paul> were), and this is not safe if the arrays contain types that
Paul> have nontrivial C++ constructors and destructors.

Yes, you are right.  Honestly, before I was perfectly able to use 1.28
on a C++ project of significant scale.  Now it no longer works because
we wanted to please Hans.  Let's stop that now.

Paul> In that case, the current bison.simple seems to be about right
Paul> for the subset of C++ that it attempts to handle.  It doesn't
Paul> work with nontrivial C++ constructors and destructors, and if
Paul> your C++ compiler works correctly you'll get a compile-time
Paul> error.  Adding support for nontrivial C++ can wait until after
Paul> the next release.

We all agree on this.  End of that topic.

Paul> That being said, I can see some simplifications of the current
Paul> copying code, which should make the code run faster in the
Paul> non-GNU case, and which apply to C as well as to C-like C++.
Paul> I'll propose these in a separate message soon.

Paul, from now on, it's the trunk only.  Work over 1.30x is over.