Designer Shape Builder bugs or oddities?

[ELinder]

I’m finding shape builder very finicky. I often get “shape builder failed” messages but no indication why. It also often does not clean up geometry properly and fails to follow the “use style from first selection”.

Attached are 2 files. The first one fails most of the time when selecting all the curves and activating shape builder, before even trying to do anything. The second does not properly add shape sections, with internal nodes not cleaned up and extra curves created and stroke and fill changed.

Intel Mac Designer 2.5.0 on MacOS version 12.7.5 (21H1222)

puzzle_pieces_shape_builder.afdesign shape-builder_falis.afdesign

[Hangman]

Hi @ELinder,

16 hours ago, ELinder said:

I often get “shape builder failed” messages but no indication why. It also often does not clean up geometry properly and fails to follow the “use style from first selection”.

Attached are 2 files. The first one fails most of the time when selecting all the curves and activating shape builder, before even trying to do anything.

Reversing the Curve Direction for the Bottom Right jigsaw piece will resolve the problem for shape-builder-fails.afdesign. I'm not entirely sure why this is necessary since the curve direction is the same for all five shapes...

The second one I'm not entirely sure why this isn't working as expected at the moment but the Shapebuilder tool seems to struggle with perfectly overlapped shapes...

[ELinder]

Wow, interesting find on needing to reverse the curve. There's definitely something wrong if it can’t figure that out on its own. All the larger puzzle pieces were initially created with shape builder from another set of intersecting curves, then I'm further working on them. I got many “shape builder failed” errors when I was creating them, so you may be right about the tool having problems with perfectly overlapping shapes. Which seems odd for a vector program where that's massively prevalent. 

[Dan C]

Thanks for your report & file provided @ELinder!

I can confirm that I've been able to replicate both the "Shape Builder Failed" error and the need to reverse the curve in the file, which I have logged with our development team.

In regards to the second file provided, I can see some interesting behaviour with the stroke width that I'm unable to recreate in a new file.
Most of the "before" curves have a stroke width of 1.5625pt, however the "bottle piece" has a stroke width of 1.529948pt.

Adding all of the puzzle pieces together using the Shape Builder, with the "Use style from first selected area" option enabled, creates a curve with a stroke width of 0.005706 pt, 0.004652pt, or 0.000006pt, which I cannot currently explain.

However, in regards to the 'internal curves' not being cleaned and the shape being unexpected, I can confirm this is technically correct behaviour. Inspecting these curves, they are not perfectly aligned with one another - meaning the tool detects each curve / overlap section in this area separately.

This can be seen when using the Freehand selection method, and ensuring to cross all of these boundaries means the curves are now correctly combined as you'd expect -

 

Therefore I will also be logging this file separately with our team regarding the stroke width changing, however the 'internal curves' either require the objects to be perfectly pixel aligned, or ensure the Shape Builder Tool has selected all of these areas before creating the shape.

I hope this helps

[ELinder]

Thank you for looking into this. However, if the main puzzle shapes are not perfectly aligned then that is also a shape builder error since they were created using it. Imagine a large grid of intersecting curves like a # sign but with more lines. I used shape builder to create all the individual pieces, all from the same source curves. During that initial process I also got numerous “shape builder failed” errors using just the source curves.

[Dan C]

1 hour ago, ELinder said:

However, if the main puzzle shapes are not perfectly aligned then that is also a shape builder error since they were created using it.

That's helpful information, many thanks for letting me know!

I'll be sure to include this within the development log & I have also noted that the shape builder tool could be 'smarter' with these specific objects and should detect these areas automatically.

1 hour ago, ELinder said:

During that initial process I also got numerous “shape builder failed” errors using just the source curves

It's possible this is related to the issue in the first file provided, though our team should be able to determine this further once they have inspected the curves in this document and hopefully resolve this for a future update 

[Hangman]

Even though the Node X and Y coordinates for the two horizontal jigsaw pieces match exactly where the two join the two pieces don't overlap perfectly... This is enough to cause the issue you see when using the Shape Builder tool...

To correct this I've split the two jigsaw pieces and rejoined them so the match is 100% perfect, this results in the correct outcome when using the Shape Builder tool... If you unselect 'Use style from first selected area' this maintains the stroke value when the pieces are merged but is this because the jigsaw pieces have a stroke with no fill...

I'm slightly unsure how this is supposed to work but when merging the shapes using the Shape Builder tool if the first selected area is the empty fill is this the style that is being matched or should that selection take into account the stroke as well?

Here you can see that selecting the empty fill as the starting point results in a merged shape with a stroke of 0.004652 pt (no idea why), but selecting the stroke before merging works as expected...