[Bug]: Offset by 1 issue for canScrollNext when tab size has fractional width

[yasuhiro-yamamoto]

The issue was caused by rounding differences between actual slide widths and offsetWidth calculations:

• When using NodeRects for size calculations, offsetWidth returns integer values even when actual slide widths have decimal points
• This integer rounding caused a 1px discrepancy in the scroll possibility check array when slide widths were fractional
• As a result, canScroll returned incorrect boolean values for certain window/slide sizes
• Added a 1px tolerance filter to prevent unexpected behavior from these sub-pixel differences

This fix ensures consistent canScroll behavior regardless of fractional slide widths by filtering out negligible 1px differences that should not affect scroll possibility determination.

• Closes: [Bug]: Offset by 1 issue for canScrollNext when tab size has fractional width #899.

[davidjerleke]

Hi @yasuhiro-yamamoto,

Thank you for your contribution. I'm working on v9 at the time of writing so it will take time before I have the chance to check this out. Thanks you for your patience.

One question: Regarding steps to reproduce the problem, can you reproduce the problem you describe simply by following the steps in this bug report?

• [Bug]: Offset by 1 issue for canScrollNext when tab size has fractional width #899

[yasuhiro-yamamoto]

@davidjerleke Thank you for your response.

Yes, I can reproduce the issue with the steps provided in the Issue. Here are the key points that I've confirmed also
reproduce the issue in playgrounds/embla-carousel-playground-react:

• When the view displays an even number of slides, the issue occurs if the total number of slides is even (e.g., with flex-basis: 50%, the issue appears with 4 or 6 total slides)
• When the view displays an odd number of slides, the issue occurs if the total number of slides is odd (e.g., with flex-basis: 33.3%, the issue appears with 5 or 7 total slides)

I noticed that my initially proposed code wasn't comprehensive enough and could still cause bugs in some cases, so I've added a commit with additional fixes.

I appreciate your time in reviewing this when possible.

[davidjerleke]

@yasuhiro-yamamoto I've added tests and made some small adjustments. I think this is good to go. Thank you a lot for your contribution ⭐! This information helped me understand the problem and write appropriate tests.

[yasuhiro-yamamoto]

Thank you for your feedback! I'm really glad I could help. This is such a great library, looking forward to contributing more!

[davidjerleke]

@yasuhiro-yamamoto a bug fix for this was just released in 8.5.2. Thank you a lot for your contribution 🎉!