Something's wrong with squeezeJoin (was: silence when using slow before chop)

[felixroos]

when slowing down a pattern before calling .chop , the pattern will return nothing:

let p = s("hh").chop(4) // this works
let p = s("hh").chop(4).slow(2) // this also works
let p = s("hh").slow(2).chop(4) // this doesnt work

repl link

Could this be a bug in _squeezeJoin ? I am not entirely sure what the expected result should be, probably the same as line 2.

[felixroos]

same with ply, which also uses squeezeJoin: https://strudel.tidalcycles.org?SB-hFm0uROHV

[yaxu]

s("hh").slow(0.5).chop(4)

is ok..

In tidal chop 4 $ slow 2 $ s("hh") seems OK but ply 4 $ slow 2 $ s("hh") is not.

In tidal ply is implemented using a squeezeJoin and chop isn't.. Which backs up your theory that there's something up with squeezeJoin.

[yaxu]

This does work in tidal:

*Sound.Tidal.Context> ply 2 $ slow 2 $ s "bd sd"
(0>½)|s: "bd"
(½>1)|s: "bd"

this doesn't

*Sound.Tidal.Context> ply 2 $ slow 4 $ s "bd sd"

works

*Sound.Tidal.Context> ply 2 $ slow 2 $ s "hh hh"
(0>½)|s: "hh"
(½>1)|s: "hh"

doesn't

*Sound.Tidal.Context> ply 2 $ slow 2 $ s "hh"

[yaxu]

Doesn't work in tidal 2.0 either..

*Sound.Tidal.Signal.Base> _ply 2 $ _slow 4 $ fastcat [sigAtom "hh" :: Signal String, sigAtom "bd"]
[]

[yaxu]

I think the problem starts here:

  _compress(b, e) {
    if (b.gt(e) || b.gt(1) || e.gt(1) || b.lt(0) || e.lt(0)) {
      return silence;
    }
    return this._fastGap(Fraction(1).div(e.sub(b)))._late(b);
  }

squeezeJoin is trying to get _compress to expand a pattern into a hap larger than a cycle, and it's refusing..

[yaxu]

Getting somewhere, not quite there yet though..

[yaxu]

Continuing from comments on #220

Ok this returns 0.5 and not 3.5, so there's definitely something wrong with squeezeJoin.. So something makes sense now.

pure(time.struct(true))._squeezeJoin().queryArc(3,4)

[yaxu]

Now I understand - the inner pattern of ply is just the original value wrapped in pure, repeating forever. So, it doesn't matter that you always get the first cycle of that. All the cycles are the same. The progression of time comes from the outer pattern.. But squeezeJoin still needs fixing for other cases like the above, where the inner pattern isn't the same every cycle.