This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: "Encapsulated" exception handling?
- From: Brad Spencer <spencer at infointeractive dot com>
- To: libstdc++ at gcc dot gnu dot org
- Date: Tue, 27 Apr 2004 10:43:01 -0300
- Subject: Re: "Encapsulated" exception handling?
- References: <20040426174335.L11414@infointeractive.com>
On Mon, Apr 26, 2004 at 05:43:35PM -0300, Brad Spencer wrote:
> Is exception handling "encapsulated" in that an object file (or shared
> library) that catches "all" exceptions can use a different version of
> g++ than other exception-using code that calls it? If stack unwinding
> never "leaves" the object file (or shared library), do the exception
> handling mechanisms of it and its callers have to be "the same"?
I later found <http://gcc.gnu.org/ml/libstdc++/2003-12/msg00071.html>,
a followup to Benjamin's response about essentially the same issue
(can the C++ ABI completely hide behind a C API). It is still not
entirely clear to me whether one of Milan's statements is "the way
things are".
He wrote:
> To me, the ABI implementation inside the library should not interfere
> with the ABI implementation outside, if the API between them is "C"
> and there are no exceptions/type_info crossing the shared object
> boundary.
Is it just that some symbols _must_ be left unhidden? Or is hiding
symbols and not passing anything C++ across the C API just not enough?
--
------------------------------------------------------------------
Brad Spencer - spencer@infointeractive.com - "It's quite nice..."
Systems Architect | InfoInterActive Corp. | A Canadian AOL Company