Jean-François Burnol submitted an update to the
xint
package.
Version number: 1.2k 2017-01-06 License type: lppl1.3
Summary description: Expandable operations on long numbers
Announcement text: ---------------------------------------------------------------------- - The xintfrac.sty floating point macros since 1.2f round their arguments to the target precision P before further processing. This rounding is now exact (aka correct) in all cases, even with fractions having long numerators and denominators.
This change has little influence on float expressions, as the xintfloatexpr parser handles there the / symbol as an operator hence it does not (except for special constructs) get to see fractions as such.
- Half-INT powers A^x (only available in float expressions, not via macros) proceed by an INT power and then a square-root extraction: the 1.2f implementation did the latter on an already rounded value, 1.2k keeps some of the guard digits to make the computed value Z closer to the exact one: a difference of less than 0.52 ulp(Z) is guaranteed in all cases.
- Macro xintnewdummy is made a public one, it serves to declare additional letters as dummy variables in expressions. This is for Unicode engines, mainly, as all Latin letters are already predefined to act as such.
See CHANGES.pdf or CHANGES.html for the details and more. ----------------------------------------------------------------------
This package is located at http://mirror.ctan.org/macros/generic/xint
More information is at http://www.ctan.org/pkg/xint
------------------------------------------------------------------------
Thanks for the upload.
For the CTAN Team Manfred Lotz
We are supported by the TeX user groups. Please join a users group; see http://www.tug.org/usergroups.html .