Agena

2.13.0, September 11, 2018 - The new function `math.stirnum` computes the Stirling number of the first or the second kind. - The new function `math.pochhammer` computes the Pochhammer function (rising factorial). - The new operator `cube` raises a number or complex number to the power of 3. - The new binary operator `roll` rotates a two-dimensional vector, represented by a complex number, through a given angle. The corresponding function `rot` with the same functionality has been deprecated. A macro has been provided to ensure backward compatibility. The operator is around ten percent faster than `rot` on Windows, and 27 percent on Solaris. - `math.clip`, `math.clamp` and `math.symtrunc` are almost equal in functionality, so all the different flavours have been combined into an extended version of `math.clip` which now also allows to user-define the return value if its argument is not in the given range. Thus, `math.clamp` and `math.symtrunc` have been deprecated. Macros, however, provide backward compatibility. - `math.fdima` has been integrated into `math.fdim`. A macro has been provided to ensure backward compatibility. - The new `bytes` package provides the userdata data structure 'double' storing an Agena number as both a C double (i.e. Agena number) and its two higher and lower 32-bit unsigned integer representations, along with functions to query and assign its individual components. - The following related `math` functions have been moved to the `bytes` package which can be activated through the `import` statement: `fpbtoint`, `gethigh`, `gethighlow`, `getlow`, `inttofpb`, `leadzeros`, `leastsigbit`, `mostsigbit`, `reverse`, `onebits`, `sethigh`, `setlow`, `tobinary`, `tobytes`, `tonumber`. - `math.sethighlowuint64` and `math.gethighlowuint64` have been removed since they could not always process each unsigned 64-bit integer correctly with Agena's doubles-based numeric system. - The following related `math` functions have been moved to the new `fastmath` package which can be activated through the `import` statement: `invroot`, `lbfast`, `reciprocal`, `sincosfast`, `sqroot`, `sqrtfast`. - In OS/2 and DOS, the `gamma` function now also processes complex numbers. - `integral` and `float` have become five percent faster. - `calc.weier`, `calc.polyfit` and `calc.polygen` have become more precise by internally using Kahan-Ozawa summation. - `math.fibinv` has not been correctly registered. This has been fixed. - The following functions deprecated one or more years ago have been finally removed: Removed function Substitute ---------------- ---------- debug.system environ.system dimension tables.dimension irem << x, y -> x symmod y >> os.curdir os.chdir() rect math.rectangular resetd stack.resetd roundf round shift << x, y -> if y > 0 then x <<< y else x >>> y fi >> - The undocumented baselib function `sunlog2` has been removed. - This release has been Valgrind-checked on x86 Linux to ensure there are no memory leaks.