[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Halfway cases for binary-to-decimal conversion in IEEE 754-1985



Date: Thu, 9 Apr 2009 15:37:45 +0200
From: Vincent Lefevre <vincent@xxxxxxxxxx>
To: stds-754@xxxxxxxxxxxxxxxxx
Subject: Re: Halfway cases for binary-to-decimal conversion in IEEE 754-1985

On 2009-04-09 05:48:11 -0700, Dan Zuras IEEE wrote:
    Look more closely at the 1985 text & you will discover that
    binary <--> decimal conversion is only required to be correct
    to 1.5 ULPs.  The reason for this is that efficient algorithms
    were not yet known at the time.

This is not what I read:

  Conversions shall be correctly rounded as specified in Section 4 for
                       ^^^^^^^^^^^^^^^^^
  operands lying within the ranges specified in Table 3.

        Since your complaint was about where it DID NOT specify
        correct rounding I presumed you were not talking about
        where it does.


    But this hardly matters because the 1985 text was obsoleted
    when the 2008 text was published.

No, this matters, because e.g. the language standards may remain
linked to IEEE 754-1985. For instance, this is the case of the
ISO C99 standard (with its annex F), which has not been revised
yet. If IEEE 754-2008 has stricter requirements than the 1985
version, a conforming C99 implementation will not magically
become non-conforming just because IEEE 754-2008 has obsoleted
the 1985 version.

-- 
Vincent Lefèvre <vincent@xxxxxxxxxx> - Web: <http://www.vinc17.org/>

        I suppose the lawyers will have to interpret this one but
        my understanding is that 754-1985 is no longer in force.
        Standards that depend on it can still be said to conform
        to it but that conformance no longer matters.  And such
        standards may have to be updated to be said to conform
        to 754-2008 but in most cases I suspect the update will
        consist of a global exchange edit of '1985' for '2008'.

        That is my understanding anyway.  Your lawyers may vary.


                                Dan


754 | revision | FAQ | references | list archive