Fw: [cxx-abi-dev] Decimal Floating Point mangling was(Fw: [cxx-abi-dev] C++0x: Mangling of rvalue reference type)s
Sean Perry
perry at ca.ibm.com
Thu Aug 16 13:44:11 UTC 2007
I think we need both mangling. We need the native type for the C
compatibility reasons Michael mentioned. The native type will be
independent from the C++ class type. We should define a mangling for the
C++ class types too. This is similar to the complex types. The ABI has Cx
for the C type and mangled forms beginning with St for the std::complex<>
classes.
--
Sean Perry
Compiler Development
IBM Canada Lab
(905)-413-6031 (tie 969-6031), fax (905)-413-4839
Dennis Handly
<dhandly at cup.hp.c
om> To
Michael Wong/Toronto/IBM at IBMCA
08/15/2007 08:50 cc
PM cxx-abi-dev at codesourcery.com,
premanand.rao at hp.com
Subject
Re: Fw: [cxx-abi-dev] Decimal
Floating Point mangling was(Fw:
[cxx-abi-dev] C++0x: Mangling of
rvalue reference type)s
>From: Michael Wong <michaelw at ca.ibm.com>
>We introduce many C99 concepts too into the C++ ABI even though they are
not
>required from the C++ Std for similar reasons.
That's fine. It just seems like we need this conversation documented.
>> And this form is mentioned:
>> ... the following catalog of abbreviations of the form "Sx" are used:
>This is to cover the case for vendors who actually wish to
>implement them as builtin types.
They can still mangled them as classes.
>From: Daveed Vandevoorde <daveed at edg.com>
>So I think Dennis is right: No special mangling code is needed here.
>If a C++ compiler has native support, it should just disguise that
>native support as a class type (from the ABI perspective, at least).
Well, the problem with a class type is the size.
We may want to use Sx to reduce the mangled string.
>From: Michael Wong <michaelw at ca.ibm.com>
>then we would need the mangling and settle it to be one or the other.
Yes. But I wanted the rationale added to the ABI, why we didn't use
the class form, or the Sx abbreviation, or some other X?
More information about the cxx-abi-dev
mailing list