Agena

agena >> `The Power of Procedural Programming` 6.6.11 Shiraz, January 30, 2026 - `getmetatable` has been extended and allows to retrieve a metatable directly from the Agena registry instead of doing so by querying a certain individual value, such as a structure or userdata. You must pass the name of the metatable and the `true` option, though, to accomplish this, for example: > getmetatable('numarray', true): [__add ~ procedure(01D6FDF8), __aeq ~ procedure(01D64150), ...] Some safeguards have been implemented to prevent abuse, see the Agena Reference. - New `long.trigamma` and `long.tetragamma` implement the trigamma and tetragamma functions for 80-bit long doubles. - Stymied error messages of `os.getconsolemode`, `os.setconsolemode` and `os.shellcolour` have been fixed. - Error messages of `ads.open`, `ads.read`, `calc.sections`, `factory.count`, `long.count`, `math.length` are no longer corrupted. - With unknown file I/O or socket errors, many Agena functions caused memory leaks when printing the error message. This has been fixed. - Added a few human-readable error messages for failed socket IO. - On Windows, `printf` when called in colour mode, issued an error if it could not determine the console mode. This can happen when stdout is redirected to NUL, especially in batch scripts. Now `printf` just prints the text without colour in these cases and no longer stops execution. - New `dual.gamma` and `dual.beta` implement the Gamma and Beta functions for (hyper-)dual numbers, respectively. - New `dual.psi` implements the Psi/Digamma function. - The `lngamma` and `invsqrt` operators now support (hyper-)dual numbers. - `dual.erfcx` now supports hyper-dual numbers. - New `dual.hypot4` computes tzhe cathetus sqrt(x^2 - y^2) for (hyper-)dual numbers. - New `dual.pytha` computes squared sum x^2 + y^2 for (hyper-)dual numbers. - New `dual.pytha4` computes squared difference x^2 - y^2 for (hyper-)dual numbers. - New `dual.totable puts all the (hyper-)dual number components into a table. - Added the new `dual.generate` function which easily allows to create new functions processing (hyper-)dual numbers. All you need are the operators and procedures working on ordinary Agena numbers [sic !], the function definition itself plus its first and second derivative. Example: > dual.lnGAMMA := # implements ln(gamma(x)) > dual.generate(<: x -> lngamma(x) :>, calc.Psi, <: v -> calc.Psi(1, v) :>, 'dual.lnGAMMA'); This allows you to utilize the whole realm of operators and functions that you find in the base, `calc`, `stats` and other libraries, to compute derivatives accurately. - The minimum CPU requirement for the Windows installers that have the phrase `-sse2-` in the file name is: Pentium 4/Athlon 64 or newer, SSE2 instruction set. AVX is not required. The SSE2-compatible Windows version of Agena is around 1.5 percent faster than the one using the x87 instruction set.