[boost] Re: typeof: automatic ID generation and GCC problems

From: Giovanni Bajo (giovannibajo_at_[hidden])
Date: 2004-07-30 20:24:45

Aaron W. LaFramboise wrote:

>> a.cpp:9: internal compiler error: in write_type, at cp/mangle.c:1555
>
> This is this bug:
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11078
>
> Apparently this is quite an old bug.  It also appears in the latest
> development version of GCC as of a few days ago, so it seems as if its
> not fixed yet.  Quite likely not to be fixed soon without someone
> showing some interest, or fixing it themselves.

No, the real point is that this is a defect in the Itanium C++ ABI, which does
not specify a way to mangle typeof (being non-standard...) nor other complex
expressions used in function signatures (like function calls or method calls
within sizeof expressions). I am told that the C++ ABI committee is reluctant
to update the ABI at the moment because there have been some talks in the C++
reflectors to prohibit such degenerated expressions.

Of course, GCC could just emit a "sorry: cannot mangle this due to a defect in
the C++ ABI" like we do for the other situations, instead of just segfaulting.
I will take care of this personally.

As for a workaround, the whole point is to bring the typeof() out of the
function signature, so that it does not need to be mangled anymore. I see
Aleksey already provided a working solution along these lines, and it looks
like the best workaround to me.
-- 
Giovanni Bajo