In article <37m728$g22@engnews2.Eng.Sun.COM>,
Steve Clamage <clamage@Eng.Sun.COM> wrote:
>As Jason Merrill pointed out in another post, the C++ standard will have
>nothing to say about name mangling. It is just one possible implementation
>technique; implementations not involving name mangling are also possible.

Good thing, too, because name mangling is but a tiny piece of what
you would need to standardize to achieve interoperability between
compilers. Personally, I think it's a feature when different
compilers use different name mangling because it makes it obvious
that they aren't compatible.

Regards,

Rob Sartin
Pencom Software

I am looking at the Function Name Encoding section (pg 122) of the ARM
which does not include any information on template encoding. The output I
see from Visual C++ is also different from the ARM.

I'd appreciate any pointers to updated information on function name
encoding and whether Visual C++ 1.5 or 2.0 are conformant.

Thanks, Jon.

>>>>> Jonathan Priddey <jpriddey@solomon.technet.sg> writes:

> I'd appreciate any pointers to updated information on function name
> encoding and whether Visual C++ 1.5 or 2.0 are conformant.

There is nothing for Visual C++ to be conformant to; many compilers use a
variant of the ARM encoding, but there is nothing in the Working Paper
about mangling.  That is too low-level to be dealt with in the language.

Jason

In article rh3@raffles.technet.sg, jpriddey@solomon.technet.sg (Jonathan Priddey) writes:
>I am looking at the Function Name Encoding section (pg 122) of the ARM
>which does not include any information on template encoding. The output I
>see from Visual C++ is also different from the ARM.
>
>I'd appreciate any pointers to updated information on function name
>encoding and whether Visual C++ 1.5 or 2.0 are conformant.

If you look carefully at that section in the ARM, you will see that it is
marked as commentary only, meaning that it is not part of the language
definition, and also that Stroustrup recommends that different implementations
use different encodings.

As Jason Merrill pointed out in another post, the C++ standard will have
nothing to say about name mangling. It is just one possible implementation
technique; implementations not involving name mangling are also possible.
---
Steve Clamage, stephen.clamage@eng.sun.com