Updating from 1.0.4 to 1.0.16 changed the appearance of the output significantly

[Julius Tarng]

Did something change in the algorithm? I bumped the package today to fix a bug (1.0.4 was generating invalid polygons in certain case) — good news is the bug is gone, but the shape has changed significantly with the same settings.

Any insight on what changed?

1.0.4:

1.0.16:

getStroke(coords)

[
  [-696, 1059],
  [-696, 1059],
  [-674, 857],
  [-607, 520],
  [-562, 116],
  [-472, -289],
  [-427, -603],
  [-404, -626],
  [-404, -536],
  [-404, -266],
  [-427, -64],
  [-427, -42],
  [-337, -177],
  [-157, -469],
  [23, -738],
  [247, -1008],
  [382, -1143],
  [405, -1165],
  [405, -1120],
  [405, -985],
  [427, -626],
  [427, -469],
  [427, -424],
  [472, -491],
  [539, -603],
  [697, -761],
  [809, -918],
  [854, -963],
  [877, -963],
  [877, -918],
  [921, -693],
  [944, -379],
  [989, -132],
  [989, 71],
  [989, 205],
  [989, 228]
];

settings:

{
              smoothing: 1,
              streamline: 1,
              thinning: 0,
              start: {
                cap: true,
                taper: 0
              },
              size: 235.883074766576
            }

[Julius Tarng]

i can get kinda close with smoothing .6 streamline .7 but it doesnt feel the same

[Julius Tarng]

the new end caps are much cleaner tho! dig it.

just concerned about regressing our users data

[Steve Ruiz]

Hey @tarngerine, this one indeed had more of a visual change than I expected on some projects. Sorry! I'll be working to make sure the lines stay more or less consistent between versions. There aren't many other changes between the two versions (primarily bug fixes related to lines with one or two points), so if the earlier version is still working for you, you may want to stay on that version.

[Julius Tarng]

Gotcha, that's what I figured, and we did have to update bc of the bug fixed. Not a huge deal! Thanks!