Re: Partial symbol export vs --export-dynamic

[Jean-Francois Panisset <panisset at discreet dot com>]

 <15149.27217.394406.651605@scooby.apac.redhat.com>Ben Elliston writes
>(The reason I mentioned C++ is that I think it's unreasonable to
> expect users to deal with mangled names--it's inconvenient.  On the
> other hand, having the user list the mangled symbols means that it's
> a little more resilient to different compilers).
>
>Ben

Isn't it the other way around? If I'm trying to distribute a multi-platform
package which may use the native compiler but the GNU linker,
and if the native compiler mangles names differently
than gcc, I'd have to distribute different "exports files" for the different
compilers, whereas if the file accepts demangled names, there should
be no such problem. Admitedly, this is a somewhat contrived example.
Right now, this is not an issue for me: although the application I'm
working on is C++, the external plugin API which I want to expose is C.

I'll be giving H.J.'s patch a try as soon as I can figure out how to
get the latest binutils CVS snapshot and build it on my machine. I'll
report back on the results.

JF