size… functions in favor of width….mag (magnitude) and mig (mignitude).The library is now organized as 4 packages:
interval_base defines the module Interval that groups the functions that work on any IEEE-754 processor and offers basic module signatures;interval_intel: defines a module Interval_intel using assembly instructions on Intel Processors;interval_crlibm: defines a module Interval_crlibm using the library CRlibm to evaluate standard functions (sometimes a bit slower but proved enclosures in contrast to the Intel package for which enclosures are not always 100% correct).interval: a meta-package that install all above three.Interval.T is a module signature to form the base of what is expected of any interval package.invx (extended inverse), cancelminus, cancelplus, inter inter_exn, low, high.equal, =, subset, <=, >=, precedes, interior, <, >, strict_precedes, disjoint.is_bounded, is_entire.I.half_pi ∋ π/2 and I.entire for [-∞, +∞].Interval_crlibm: functions expm1, log1p, log2, log10, cospi, sinpi, tanpi, acospi, asinpi, atanpi, not present in the Intel version.I.U also restores inequality relations.Interval library by using sub-modules and standard mathematical names. In particular, all operations — including infix operators — are in a sub-module I which can conveniently be used to introduce local scopes after issuing open Interval.Division_by_zero and Domain_error that are local to Interval.Fpu module has been redesigned: the rounding up or down of functions is controlled by the sub-module (Low or High) to which they belong. This allows for natural expressions such as Low.(x**2. +. 2. *. x +. 1.).