Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: roundTiesToEven in precision 1



> To:   Vincent Lefevre <vincent@xxxxxxxxxx>
> Cc:   "stds-1788-er" <stds-1788-er@xxxxxxxxxxxxxxxxxxxxx>,
>       "stds-1788" <stds-1788@xxxxxxxxxxxxxxxxx>,
>       Dan Zuras Intervals <intervals08@xxxxxxxxxxxxxx>

Now you're posting to 1788 what should have gone to 754!

> ... we must have p > 1 & emax >= 1.

Didn't you like my explanation that the rounding rule worked just
fine for p=1, if we add a statement about rounding overflow, which
only matters when precision=1?

Note that clause 5.12 explicitly supports a requested output
precision of 1 digit, for both hex and decimal, so we cannot
now change our mind and prohibit p=1 -- except possibly for
the "extendable" formats.  (The "extended" formats are ok
since their p must exceed that of the basic formats already.)

Michel.
Sent: 2009-03-25 23:18:59 UTC