* rename `beat` option to `stackBy` to `repeat`
* fix parse error reporting
* rename `weight` to `tactus` (it might in the end be pulse, step, or tap)
* tactus marking with ^
* and add some tests
- `.` feet now work, e.g. `"a b c . d e . f"` is the same as `"[a b c] [d e] [f]"`
- `_` elongation now works, e.g. `"a _ _ b"` is the same as `"a@3 b"` (`"a @ @ b"` also works, like in tidal)
- standalone `!` now works, e.g. `"a ! ! b"` is the same as `"a!3 b"`
Reworks #47 and #49
Ref #30
- transpiler now uses m function with globalOffset
- patternifyAST now accepts global offset
- patternifyAST now calls .withLoc with global leaf location
- .withLoc replaces .withLocation + .withMiniLocation
- simple locs (offsets) are now used everywhere
- some tests fail, seems some haps have reordered...
- wip: Repl still uses hardcoded updateMiniLocations
- todo: find way to call updateMiniLocations dynamically
fixes#504
* support list syntax in mininotation
* support compound controls
* remove redundant : splitting of s/n and note/n from webaudio
* patternable scale names
* port Rohan Drape's Bjorklund implementation, add Toussaint's tests
* fix euclidLegato, simplifying a bit now that bjork results should always begin with an 'on'
* migrate euclid numbers in tunes
- 3,4 +1
- 5,8 -1
- 6,8 +3
Co-authored-by: Felix Roos <flix91@gmail.com>
* patternify the mininotation bjorklund, and fast (*) slow (/) and pattern step (%) parameters
* simplify replicate (!), should be closer to tidal now
* highlight atoms rather than elements
* tests
* rename parser stuff to be more explicit and fit with tidal concepts. qualify all strudel function calls
* Add {,}%n polymeter support, with a few tests
* remove _ prefixes except for functions to be patternified
* categorise pattern methods
* experimental support for `.add.squeeze` and friends as alternative to `.addSqueeze`
* `every` is now an alias for `firstOf` with additional `lastOf` (which every will become an alias for next)
The tests are probabilistic, so it is possible that if the
pseudo-random number generator changes in the future, we might
get results that fail. They work for the current PRNG, though,
and use boundaries for the number of values of different types
such that there should only be about a 1% probability that the
tests would fail by chance assuming that the PRNG returns
evenly distributed values.
