Contributors:
- Ben Elliston
- Janis Johnson
- Yao Qi
Dwayne McConnell
Delivery Date:
- Unknown
Description:
- The IBM Linux Technology Centre is working to add the C extension for decimal floating point arithmetic into GCC. See This is an existing ISO/IEC working group (WG14) project. We are currently developing this on a branch of the FSF repository and hope to include it in the GCC 4.2 release. I am unsure of the status of the draft and when it might be ratified into the ISO C standard. It seems that the corresponding changes to the C++ standard are a little further behind, so we will not be attempting C++ support at this stage. For certain applications, particularly transaction systems that handle decimal arithmetic, these changes will allow developers of such applications to write to a single language standard instead of relying on any number of existing and differing (syntactically and semantically!) software libraries. We will be providing the required changes to the compiler and to the runtime. As the draft calls for additional library functions, we will be providing separate patches to glibc. At this stage, we are deliberating over whether a machine-independent feature like this should be developed within GCC generically (higher impact) or whether we should keep it as a change that is only applicable to IBM targets (makes less sense, but minimises impact).