Comment #16 on issue 436 by [email protected]: Numeric results are
wrong on some systems due to x87 extended double precision
https://code.google.com/p/v8/issues/detail?id=436
PLEASE READ MY ENTIRE POSTING BEFORE YOU MAKE UP YOUR MIND OR YOU WILL NOT
UNDERSTAND THE UNDERLYING ISSUE, WHICH IS JUST AS GRAVE AS THE "GOTO
CONSIDERED HARMFUL" ISSUE
IEEE 754 double extended has 80 bits internal calculation accuracy (not the
final result that is meant to be stored and/or displayed) to prevent as
much as possible bad results coming from NUMERICAL INSTABILITY due to
floating point roundoff in math formulas that is HARD TO NOTICE and in
practice may produce catastrophic ERRORRS that are SEEN TO LATE, NEVER OR
MISLABELED, for example medical imaging errors that never get blamed on the
initial code from the medical device manufacturer. EcmaScript and/or V8 may
be used for serious scientific or engineering applications, now or in the
future, that might be affected by this issue.
See http://www.cs.berkeley.edu/~wkahan/JAVAhurt.pdf page 62 (start with the
how and why headers) for a small SUMMARY FROM AN EXPERT with decades of
experience in reasoning about and debugging numerical computations, that
participated in the creation of the IEEE 754 floating point standard, and
who created that paper, and countless others accesable from his home page,
about the miriad of ways in which full adearence by EcmaScript to IEEE 754
is BEST FOR code that gets created by PROGRAMMERS WITHOUT NUMERICAL
EXPERTISE.
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
