Whatever happened to 754-1985 traps and wrapped exponents?

754-1985 has a clause 8, Traps, and clause 7 refers to
wrapped exponents for results when traps are enabled for
overflow or underflow.  The wrapped exponents are useful
for computing products and quotients that might overflow or
underflow at intermediate stages, but still produce a
result that is in range: software keeps count of how many
wraps occur, and if the number of overflows equals the
number of underflows, the computed result is contaminated
only by normal roundoff.  This situation arises, for
instance, while computing Clebsch-Gordan coefficients,
which involve quotients of products of factorials:

 https://en.wikipedia.org/wiki/Clebsch%E2%80%93Gordan_coefficients

For 754-1985's binary formats,
the exponent wrap is 3 * 2^(n-2) for an n-bit exponent field,
to map ordinary overflows and underflows back toward the
middle of the normalized range.

754-2008 does not mention traps or wrapped exponents, and
has a clause 8 Alternate exception handling attributes.
What happened?

754-1985 was concerned with getting the standard adopted
at least at a hardware level, hoping languages would follow.
To that end, it required or recommended what could be
called implementation techniques at a hardware level
that could be used to provide useful facilities at a
language level.      754-2008, in contrast, could assume
fairly widespread acceptance of 754-1985 at a hardware
level, and concentrate on the more important problem
of useful language features for portable programming.

Implementation experience with 754-1985 confirmed that it
was not possible to write a portable trap handler.    There
are too many hardware and system software constraints that
a trap handler has to be aware of.  A commonly useful
feature, often provided by compiler command line options,
is to be able to abort the computation when an unplanned
exception arises.  abort is a common feature of language
run times; but fix-and-continue trap handling (of the kind
envisioned when exponent wrapping was specified) is not.

So instead, 754-2008 sought to standardize facilities at
the level that could be useful in language standards and
exploited portably by computational applications.
Hardware traps and wrapped exponents can be useful
implementation techniques but it is the higher-level
facilities that are important and portable.

So 754-2008 clause 8 recommends that languages specify
alternate means of exception handling and suggests a number
of alternates.     It's a recommendation rather than a
requirement, like 754-1985 clause 8.    Both abort and
fix-and-continue facilities are recommended.
A hardware facility to trap on exceptions is envisioned
for performance, but the clause 8 recommendations can be 
implemented almost entirely with flags. 

As for wrapped exponents, the higher-level construct is the
scaledProd reduction operations in 9.4.    They specify
that products be computed as an implicit product of a
scaled product and a scaling factor, a power of the
radix.    That suffices for applications like
Clebsch-Gordan coefficients.     Hardware that provides a
wrapped-exponent capability on overflow or underflow can be
exploited to provide efficient evaluation of such
products.    But the reductions can be implemented without
wrapped-exponent trapping by using logB and scaleB.