Reorganise pattern.mjs with a 'toplevel first' regime

[yaxu]

No description provided.

[felixroos]

looks promising!

[yaxu]

pattern.mjs is looking tidier now. Other modules are still using patternified and/or define rather than the new register, and should probably be updated.

[yaxu]

I had a go at tidying up degrade/undegrade in signal.mjs along the same lines but hit the complication that register makes heavy use of reify, which applies pure which is for discrete values.. this messes things up.
So I propose merging for now and doing further tidying later.

[felixroos]

I found a way to get rid of the arity switch case in register! I'd be fine with a merge. We can refactor makeComposable / patternifyX / Pattern.prototype.define in a separate PR

[yaxu]

Heh, let me know your t-shirt size and address !

[felixroos]

EDIT: hm it seems the tests now fail.. dunno what went right there at first... although it sounds right. maybe randomness just behaves a little different now, similar to how haps have reordered a little bit?

I had a go at tidying up degrade/undegrade in signal.mjs along the same lines but hit the complication that register makes heavy use of reify, which applies pure which is for discrete values.. this messes things up.
So I propose merging for now and doing further tidying later.

just tried using register with degrade(By) / undegrade(By) and it seems to work even with reify. see 09b15a0

should I just refactor the rest of signal.mjs similar to this or is there something messed up I am missing? I also tried to add a noReify option, but skipping reify actually breaks stuff

[yaxu]

Yeah I tried the same and think the use of pure was putting a sample and hold on the random signal so it was either all or nothing on a per cycle basis.

[yaxu]

We need to reify but with steady rather than pure.

[felixroos]

I wonder why it worked with the old patternify function

[yaxu]

I'm afk but it's using something custom with applicative.. hmm