[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
Re: motion for Sep 7: change requirement to detect tininess in the same way to a recommendation
Item 15. - Dan
Date: Mon, 28 Aug 2006 13:51:11 -0700
From: Jim Thomas <jwthomas@xxxxxxxxxx>
Subject: motion for Sep 7: change requirement to detect tininess in the same
way
to a recommendation
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Dan, I'd like to make the following motion at the September 7 meeting.
-Jim
-----------------
Motion: change requirement to detect tininess in the same way to a
recommendation
Rationale: The current draft in 9.5 paragraph 1 says: "The implementer
may choose how tininess is detected, but shall detect tininess in the
same way for all operations." The standard allows two ways of detecting
tininess because it gives implementers some flexibility and because the
difference -- whether a value right at the underflow threshold signals
unnecessarily -- doesn't matter to users. I'm unaware of any argument
that the consistency requirement matters to users either. As the number
of operations required by the standard has increased, particularly for
implementers providing both binary and decimal arithmetic, there is
greater opportunity to leverage implementation across platforms. The
requirement that new operations must match existing ones in how tininess
is detected would be an unnecessary impediment to porting implementation
of new features and to developing libraries that provide standard
functionality.
Text change:
Current: The implementer may choose how tininess is detected, but shall
detect tininess in the same way for all operations.
Proposed: The implementer may choose how tininess is detected, but
should detect tininess in the same way for all operations of a given
radix.