[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
Re: 754R vs pow
to normative (optional per function) and expanded substantially. Under
discussion now are farther reaching substantive changes, for
reproducibility, that go well beyond what the working group concluded is
reasonable.
The last working group draft of 2006-09-17 said:
* For operations specified in the normative part of this standard, numerical
results and exceptions are uniquely determined by the values of the input data,
sequence of operations, and destination formats, all under user control.
* Languages that aspire toward reproducible results on all platforms are
expected to specify more behaviors than languages that aspire toward maximum
performance on all platforms.
* Languages should provide means for programmers to specify reproducible
results-
results that are identical on all platforms supporting that language and this
standard, for operations completely specified by this standard.
I think most of the discussion has been to elaborate what these statements
mean, rather than 754's inexplicable and unexplained:
* When restricted to a declared
subset of the standard, these programs should produce identical
results on all conforming systems.
These changes would be made without the breadth of
experience and open representation and process of the working group,
The sponsor ballot group is larger and is supposed to be more diverse
than the active membership of the working group. And the stds-754 list is
much larger and more diverse than either.
For the next ballot, voters might no
longer assume that the draft is, as it claims, "a product of the
Floating-Point Working Group".
The working group timed out while still making frequent significant changes
and as far as I can tell would have continued to thrash for as long as it
was allowed to. It's up to the sponsor ballot group to find some
different process to arrive at maximum consensus on how to achieve
the goals of the PAR and closure. If a final product is produced that
is too far off base, then there are institutional procedures to deal with that.