Cycseq branch - backward compatibility

yaxu commented 2 years ago

TODO

[x] chew
[x] cycleChoose
[ ] deconstruct - implement in Show?
[ ] every' (everyBack?)
[ ] fastSqueeze
[ ] fit
[ ] fit'
[x] foldEvery (better name?)
[ ] ghost
[ ] ghost'
[ ] ghost''
[x] grain
[ ] interpolate
[ ] interpolateIn (?)
[ ] jux
[ ] jux'
[ ] jux4
[ ] juxBy
[ ] juxcut
[ ] juxcut' (these need sorting out)
[ ] layer?
[ ] mono
[ ] permstep
[ ] pick
[ ] pickF
[ ] plyWith
[ ] randArcs (?)
[ ] randStruct (?)
[ ] randrun (?)
[ ] rangex
[ ] sameDur (?)
[ ] scramble (for random)
[ ] select (rationalise with pick / pickF?)
[ ] selectF
[ ] sew
[ ] shuffle (for random)
[ ] silent (?)
[ ] slowCat
[ ] slowSqueeze
[ ] slowappend
[ ] slowspread
[ ] slowstripe
[ ] smooth
[ ] spaceOut
[ ] spread
[ ] spread'
[ ] spreadChoose
[ ] spreadf
[ ] spreadr
[ ] squeeze (?)
[ ] squeezeJoinUp (?)
[ ] stackwith (?)
[ ] step
[ ] step'
[ ] steps
[ ] stitch
[ ] stretch
[ ] stripe
[ ] substruct
[ ] swap (?)
[ ] swing
[ ] swingBy
[ ] tabby
[ ] timecat (?)
[ ] toScale / toScale (?)
[ ] uncollectEvent / uncollectEvents
[ ] ut (reimplement with sequence ?)
[ ] wait/waitT (transition ?)
[ ] whenmod
[ ] within / within'
[ ] draw/drawLine/drawLineSz

From 'simple' module ?

[ ] higher
[ ] louder
[ ] lower
[ ] quieter
[ ] slower
[ ] faster
[ ] crunch
[ ] scratch
[ ] skip
[ ] left (?)
[ ] right (?)

Move to Sequence

[ ] lindenmayer
[ ] lindenmayberI
[ ] markovPat
[ ] runMarkov
[ ] seqP / seqPLoop - reimplement with sequence

Removed

stut
offadd
samples
samples'
chunk' - use chunk with negative number

Renamed

iter' -> iterBack
collectBy -> _collectBy
collectEvent -> _collectEvent
collectEventsBy -> _collectEventsBy

yaxu commented 2 years ago

@dktr0 I've been working on some major changes in the cycseq branch, towards tidal version 2.0, that have broken tidal-parse. I think things can be mostly backwardly compatible in terms of end-user code but the internals have changed quite a lot.

For example Pattern is now a typeclass, and the datatype formerly known as Pattern is now called Signal that implements it. This makes way for Sequence as a second instance of Pattern with a different representation oriented around sequences, which could be seen as a formal representation for the mininotation that will allow a lot more flexibility.

There's also a lot of added stuff, for example much more flexible strategies for aligning patterns. I can go into more details if you like.

So the question is what to do about tidal-parse. I think easier for now would be to get it compiling as-is. I.e. so tidal-parse still parses tidal 1.9.x syntax and behaviour and nothing more. Then if/when tidal-2.0 is released we can look at adding the additional operators and worry about this new signal type.

Does that sound good? Would you be interested in/have some time spare to help with the changes (if needed, it might turn out super simple, not sure..) and trying it out in estuary?

dktr0 commented 2 years ago

@yaxu sounds good! If not too much changes about the "surface" syntax of tidal-parse, it probably is fairly straightforward - there's not a lot of exposure to Tidal's "internals" in tidal-parse. Is tidal-parse still working okay with 1.9.x?

dktr0 commented 2 years ago

(I guess a lot of type signatures will change...)

tidalcycles / Tidal