[Cfp-interest] WG14 IEEE 754-C binding meeting minutes 2017/09/26
Rajan Bhakta
rbhakta at us.ibm.com
Tue Sep 26 11:05:16 PDT 2017
Attendees: Rajan, Jim, Fred, Mike, David H., David C.
New agenda items:
None.
Last meeting action items:
Rajan: Check with David Keaton to see if SD3 will be considered in the
next WG14 meeting and if so, can the floating point proposals (other than
part 1 and 2) be postponed unless we are prepared for them. - Done.
Jim: List the proposals we have in flight with WG14 and what we need
to do with them (if anything). - Done.
Jim: Create a new DR against part 3 for the non-arithmetic interchange
formats to create format_DECIMAL_DIG type macros. - Done.
Jim: Augmented add tgmath: Change "invokes a real function" to
"invokes a function returning a real type". - Done.
Jim: Augmented add: Create a proposal for the new augmented precision
functions including tgmath for IEEE 2018 C binding. - Done.
Jim: Create a note to say the existing fmin/fmax functions may not
correspond to the new IEEE 754:2018 f{min/max}* functions in TS Part 1 as
part of the 2018 binding update. - Done.
New action items:
Jim: Activities in progress: Obsolesce DECIMAL_DIG: No proposal needed
as our last one was accepted (N2108).
Jim: Activities in progress: %a for decimal differing from binary:
Change "paper" to Reflector message 14792.
Jim: Activities in progress: Change all TS Part actions to "update and
re-present" instead of just re-present.
Jim: Get document number and submit DECIMAL_DIG and DIG macros for
non-arithmetic formats paper for the WG14 mailing.
Jim: Write a DR for the _Roundwise in cbrt sample for the WG14
mailing.
David H, Mike: Make a list of what has changed or been augmented in
IEEE 754:2018.
Jim: Augmented arithmetic: F.10.13 is missing the paragraph number in
the second paragraph.
Jim: Augmented arithmetic: F.10.13.3#4: computation t underflows ->
computation of t underflows.
Jim: Min/Max functions: Mention that the existing fmin/fmax functions
do not distinguish bewteen +0 and -0.
Jim: Draft a response to Willem's WG14 paper (N2166) for next meeting.
Next Meetings:
Tuesday October 17th, 2017, 12:00 EDT, 9:00 PST
Same teleconference number.
Discussion:
IEEE 754 revision:
Discussion on the payload get and set functions. Not yet resolved.
Not clear the best choice for rounding for the augmented precision
functions. Most important for higher precision. If there are other use
cases, it will help knowing them.
C++ liaison:
No update.
Activities (http://wiki.edg.com/pub/CFP/WebHome/in_flight-20170916.pdf
):
*Jim: Activities in progress: Obsolesce DECIMAL_DIG: No proposal
needed as our last one was accepted (N2108)
*Jim: Activities in progress: %a for decimal differing from binary:
Change "paper" to Reflector message 14792.
IEEE-754:2018 update: Can start on draft of a paper, but don't send
it in until it passes ballot.
*Jim: Activities in progress: Change all part actions to update and
represent instead of just represent.
For next WG14 meeting:
Obsolesce DECIMAL_DIG:
Discussed already.
DECIMAL_DIG and DIG macros for non-arithmetic formats:
Waiting for the document #. Should be in the mailing.
*Jim: Get document number and submit DECIMAL_DIG and DIG macros
for non-arithmetic formats paper for the WG14 mailing.
%a formatting consistency:
Discussed.
_Roundwise in cbrt sample (Joseph Myers):
*Jim: Write a DR for the _Roundwise in cbrt sample for the WG14
mailing.
Binding for IEEE 754-2018:
WG14 paper about updating to IEEE 754-2018:
*David H: Can make a list of what has changed or been augmented
with Mike. Will need a WG14 person to translate it into WG14 wording.
Functions for augmented arithmetic (
http://wiki.edg.com/pub/CFP/WebHome/augop_spec-20170915.pdf):
Part 4 update.
*Jim: Augmented arithmetic: F.10.13 is missing the paragraph
number in the second paragraph.
*Jim: Augmented arithmetic: F.10.13.3#4: computation t underflows
-> computation of t underflows
Should we reconsider this being IEEE only?
Any other implementations have non-IEEE and are up to date with
the standards?
Can do general wording with no specific rounding in the main
body of the standard and then specify the rounding method in Annex F.
Will wait for 754 to decide what they will do with these
functions.
Min/max functions (
http://wiki.edg.com/pub/CFP/WebHome/min-max_spec-20170924.pdf):
*Existing fmin/fmax don't distinguish between +0 and -0.
Payload functions:
If IEEE removes these functions, should they be removed from Part
1?
Mike: Decimal has the payload well defined. N-1 digits. How the
payload is used is undefined.
Jim: Implementations can use some of the bits in the payload for
themselves while some bits can be for the user. The definition of the
payload is for diagnostic information in the glossary in the 754 spec.
Not clear if the user has access to everything in the payload due
to the wording implying diagnostic information in there as well.
Don't want to conflict with any hardware or software payload
values if the user wants to set their own.
Will wait to see what 754 does.
Other:
Willem Wakker evaluation formats (N2166):
Problem 1) Agree that evaluation format should be defined.
Not sure about where the text goes (language vs environment).
Seems to fit better where it is as part of the floating point
model.
Problem 2) The change in DR500 seems to be fine and it is not
unclear.
Characterization vs defined: Due to the use of characterization
when discussing a number of things relating to this in the standard (like
number of digits).
Problem 3) Can see the potential problem, but not sure if it is
characterized correctly with the assumptions given. Ex. sloppy
programmers.
Note, sloppy programmers are the vast majority of programmers.
Problem 4) Having stricter behaviour in Annex F allows more
portability and hence it does deal with IEEE.
There was discussion in WG14 for this and it was decided to
allow wide, but was OK with restricting it in Annex F.
Don't want to change Annex F.
Fred: Some implementations do have wider return values for some
types and not for all.
Jim: Don't think we ever intended for FLT_EVAL_METHOD to affect
function returns.
Fred will follow up with clarifying this return value issue.
Jim: Need more words to get implicit conversions included with
what is in 5.2.4.2.2#9 after DR500.
TC's:
6) Looks good in principle, but should not require a cast (cast
is a runtime operation) so you need to use a static constant or hex
floating point.
*Jim: Draft a response to Willem's WG14 paper (N2166) for next
meeting.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.oakapple.net/pipermail/cfp-interest/attachments/20170926/1d5a6554/attachment.html
More information about the Cfp-interest
mailing list