[Cfp-interest] Summary of the ISO C Standards meeting presentation of IEEE 754-2008 C Binding part 1
Rajan Bhakta
rbhakta at ca.ibm.com
Tue Feb 21 11:16:47 PST 2012
N1591 (updated) - IEEE 754-2008 C binding, Monday, 2012/02/13, 11:05 am,
Jim Thomas presenting in the Spring 2012 ISO JTC1/SC22/WG14 meeting.
Preamble:
IEEE 754-2008 It is an ISO/IEEE/IEC standard. IEC 60559:2011.
It will be a single document with 5 separate parts.
Issue discussions:
Issue 1: Free standing vs hosted (Clause 2, p7)
- Can do something like stdint and inttypes, one being a subset of
another.
Ex. We can make the Floating Point aspects into a special subset that
could be used to make it easy for freestanding implementations.
Consensus to move forward to allow freestanding implementations
Note: Most people in MISRA would support this
Issue 2: Should we add a WANT macro?
- Should do this to avoid namespace pollution.
Consensus was to add this and to use a number of separate macros.
Issue 3: Table 1: How can it be improved?
- Put forward references to the section in the C Standard where the
details are into a third column. Note: A particular function may not be
completely referenceable but that is also interesting.
Consensus was to replace F.3 with the table with the extra column(s) with
the forward reference(s).
Issue 4: Withdrawn to bring up later in a different part (alternate
exception handling).
Issue 5: Call made through function pointers - All function calls are
function pointer calls
- It could be a function that returns a function pointer.
- Do something like tgmath where you just specify the function and the
correct rounded type of function is chosen.
- Whatever solution has to consider inlining.
Still open.
Issue 6: Integer width macros - Is this where this belongs? Could be
outside the Floating Point TS.
Consensus was to keep it here as part of the floating point TS.
Issue 7: Withdrawn - We will work on this: Names are not quite perfect.
Issue 8: pg. 29: Should the 7.12 spec of llogb require a domain error for
finite out of range cases?
Keep it the same as ilogb unless it changes based on Fred's papers. (Note:
Fred's papers resulted in no change at this time.)
Other comments:
It is hard to tell where the text replacement starts and ends. Can
it be clarified?
Ex. Page 10
Suggestion: Put the old text in a box with the replacement text in
another box.
Note: Cannot be published with strikethroughs, bold for inserts,
etc. but can be seen as that way for presentation. A good example of how
to do it is the embedded TR and the DFP TR.
Next steps:
1) Continue as we're working
2) Create a new work item for WG14 - Jim and John Benito to work on this.
WG14 can issue the document as each national body says they don't have a
document.
Consensus was that this is ready to be a work item for WG14.
Regards,
Rajan Bhakta
z/OS XL C/C++ Compiler Technical Architect
ISO C Standards Representative for Canada
C Compiler Development
Contact: rbhakta at ca.ibm.com, Rajan Bhakta/Toronto/IBM
Telephone: (905) 413-3995
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.oakapple.net/pipermail/cfp-interest/attachments/20120221/e7c02714/attachment.html
More information about the Cfp-interest
mailing list