From 3f972afb3aba25a0871758419be5cfef7c2c9138 Mon Sep 17 00:00:00 2001 From: Lumi Pakkanen Date: Mon, 13 May 2024 08:42:07 +0300 Subject: [PATCH] Expand technical docs --- documentation/technical.md | 98 +++++++++++++++++++++++++++++++++----- 1 file changed, 85 insertions(+), 13 deletions(-) diff --git a/documentation/technical.md b/documentation/technical.md index cd7c3351..b944c381 100644 --- a/documentation/technical.md +++ b/documentation/technical.md @@ -1,6 +1,16 @@ # SonicWeave - Technical overview This documentation describes the SonicWeave DSL as it relates to other programming languages. +## Purpose +SonicWeave is designed for notating microtonal scales as an extension of Scala .scl syntax. Programming is secondary so you'll have to dodge around reserved patterns such as `C4` which look like identifiers but correspond to musical literals. + +### Other pitfals +Fraction slash `/` binds stronger than exponentiation. Use `÷` or `%` if you need division that follows [PEMDAS](https://en.wikipedia.org/wiki/Order_of_operations). + +The exponent is required in decimal literals. `1.23e0` instead of `1.23` which is instead interpreted a musical interval 1.23 cents wide. + +The meaning of `*` changes depending on the operands. Use `~*` to mean musical stacking of relative intervals i.e. mathematic multiplication. + ## Type system Values in SonicWeave fall into these categories @@ -35,7 +45,7 @@ All operations are left-associative except exponentiation, recipropower, and log | `{x, key: value, "third key": value, ...}` | Record display | | `x[index]`, `x[start..end]`, `x[s,next..end]` | Access, slice | | `x(arguments...)` | Call | -| `interval color label` | Labeling | +| `interval color label` | Intrinsic call e.g. painting or labeling | | `++x`, `--x`, `+x`, `^x`, `∧x`, `∨x`, `/x`, `\x` | Increment, decrement, no-op, up, down, lift, drop | | `/` | Fraction | | `^`, `^/`, `/^`, `/_` | Exponentiation, recipropower, logdivision* | @@ -50,36 +60,98 @@ All operations are left-associative except exponentiation, recipropower, and log | `not x`, `vnot x` | Boolean not, vector not | | `and`, `vand` | Boolean and, vector and | | `or`, `vor`, `al` | Boolean or, vector or, niente coalescing | -| `x if y else z` | Ternary conditional | +| `x if y else z`, `x where y else z` | Ternary conditional, vector ternary conditional | | `lest` | Fallback[^1] | Parenthesis, `^`, `×`, `÷`, `+`, `-` follow [PEMDAS](https://en.wikipedia.org/wiki/Order_of_operations). The fraction slash `/` represents vertically aligned fractions similar to `$\frac{3}{2}^\frac{1}{2}$` in LaTeX e.g. `3/2 ^ 1/2` evaluates to `sqrt(3 ÷ 2)`. [^1]: `lest` is a fully associative operation, thus `a() lest b() lest c()` = `(a() lest b()) lest c()` = `a() lest (b() lest c())`. This means that one can always treat a sequence `a() lest b() lest ... lest c()` as if the evaluations occurred from left to right. Any possible parenthesis may be ignored in such a sequence. +## Control flow + +### For..of +The contents of arrays and records can be iterated over using +```c +for (const element of container) { + /* body of the loop utilizing element */ +} +``` + +### For..in +The indices of arrays or keys of records can be iterated over using +```c +for (const element in container) { + /* body of the loop utilizing element */ +} +``` + +### While +The body of a while loop is executed until the condition becomes falsy. +```c +let i = 10; +while (--i) + i; +// Result is numbers from 9 to 1 pushed onto the implicit array $. +``` + +### Break, continue +Loops terminate on `break` and continue with the next iteration on `continue`. + +### For/while...else +The `else` branch of a for or while loop is executed unless a `break` statement was encountered. + +## Functions +Functions are constructed using either the `riff` keyword or the `fn` alias. The return value is indicated using the `return` keyword. If the end of the function body is encountered the return value is the array of intervals pushed onto the current scale i.e. `return $`. + +## Exceptions +Exceptions (strings) are raised using `throw` and handled inside `try..catch` blocks or inline using `x() lest y()`. + +## Conditional execution +Use chained `if..else`. There is no `elif`. + +## Deferred execution +To defer execution to the end of the current block prefix the statement with `defer`. Multiple statements with `defer` inside a single block are executed in reverse order. + ## Interval subtypes -TODO: List all interval subtypes +| Type | Examples | Domain | Echelon | Notes | +| ------------ | ----------------------- | ------------- | --------- | ----- | +| Integer | `2`, `5` | Linear | Relative | Same as `2/1` or `5/1`. | +| Decimal | `1.2e`, `14e-1` | Linear | Relative | Decimal commas (`1,2`) only work on isolated lines. | +| Fraction | `4/3`, `10/7` | Linear | Relative | The fraction slash binds stronger than exponentiation | +| N-of-EDO | `1\5`, `7\12` | Logarithmic | Relative | `n\m` means `n` steps of `m` equal divisions of the octave `2/1`. | +| N-of-EDJI | `9\13<3>`, `2\5<3/2>` | Logarithmic | Relative | `n\m

` means `n` steps of `m` equal divisions of the ratio `p/q`. | +| Step | `7°` | Logarithmic | Relative | Correspond to edo-steps when tempering is applied. | +| Cents | `701.955`, `100c` | Logarithmic | Relative | One centisemitone `1.0` is equal to `1\1200`. | +| Monzo | `[-4 4 -1>`, `[1,-1/2>` | Logarithmic | Relative | Also known as prime count vectors. Each component is an exponent of a prime number factor. | +| FJS | `P5`, `M3^5` | Logarithmic | Relative | [Functional Just System](https://en.xen.wiki/w/Functional_Just_System) | +| TAMNAMS | `P0ms`, `m4ms` | Logarithmic | Relative | Requires a `MOS` declaration. | +| Frequency | `440 Hz`, `2.2 kHz` | Linear | Absolute | Absolute frequency of oscillation. | +| Duration | `1 ms` | Linear | Absolute | Absolute period of oscillation. | +| Absolute FJS | `C4`, `Eb_5` | Logarithmic | Absolute* | Absolute version of [FJS](https://en.xen.wiki/w/Functional_Just_System). | +| Diamond-mos | `J&4`, `M@3` | Logarithmic | Absolute* | Absolute counterpart to TAMNAMS. Requires `MOS` declaration. | +| S-expression | `S8`, `S5..8` | Logarithmic | Relative | Additive spelling of [square superparticulars](https://en.xen.wiki/w/Square_superparticular). | +| Val | `<12, 19, 28]` | Cologarithmic | Relative | Used to temper scales. | +| Warts | `17c@`, `29@2.3.13/5` | Cologarithmic | Relative | [Shorthand](https://en.xen.wiki/w/Val#Shorthand_notation) for vals. | +| SOV | `17[^5]@` | Cologarithmic | Relative | [Shorthand](https://en.xen.wiki/w/Val#Sparse_Offset_Val_notationn) for vals. | +| Jorp | `€` | Cologarithmic | Relative | `<1200]` | +| Pilcrowspoob | `¶` | Cologarithmic | Absolute | `withEquave(<1]@Hz, 1 Hz)` | ## Domains -TODO: List domains of literals and domains of operations +Linear domain values add as in mathematics. Logarithmic domain values add by multiplying the underlying values. Modulo (`mod`) can be implemented using repeated subtraction so it turns into repeated division in the logarithmic domain. This applies to most other operators that can be reduced to addition and negation. + +The cologarithmic domain has the semantics of the co-domain of the logarithmic as a vector space where the exponents of prime numbers form the basis. ## Echelons -TODO: List echelons of literals and echelons of operations. +Anything that can be normalized to a frequency by inverting and negating its time exponent is in the absolute echelon the exponent of the Hertz unit works as a weighting factor in addition. Scalars with a zero time exponent are in the relative echelon with no hidden weights or projection. ## Ranges of values -TODO: Explain the limits of radical values and reals. - -## Control flow - -TODO: For and while loops - -## TODO +The number of prime components is set by calling `numComponents(n)`. Any value with a lower nth prime limit will then support exact square roots and other radicals. The maximum exponent of a prime number is `9007199254740991` and the same `2^53 - 1` limit applies to the denominator of the exponent. The multiplicative higher prime residue has the same range but the `Number.MAX_SAFE_INTEGER` limit applies directly to the numerator and denominator of the residual and no square root of a higher prime is representable as a radical. -TODO: Other technical notes +Real values are double precision floating point numbers. ## Next steps