[CSS Regions][CSS Shapes] ASSERTION FAILED: m_segmentRanges.size() < m_segments.size()

[Zoltan Horvath]

If we have an e.g. up-side-down triangle, then at the bottom part of the triangle the content doesn't fit, then it needs to be adjusted. The adjustment to the next region happens correctly, but the current region/shape doesn't get updated, which leads to the the shape mismatch, which leads to layout error / assertion.

[Zoltan Horvath]

Created attachment 219304 [details]
Patch

[Darin Adler]

Comment on attachment 219304 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=219304&action=review

> Source/WebCore/rendering/RenderBlockLineLayout.cpp:1231
> +        RenderRegion* regionAtNewLogicalHeight = regionAtBlockOffset(logicalHeight());
> +        if (!currentRegion->isLastRegion() && currentRegion != regionAtNewLogicalHeight) {
> +            currentRegion = regionAtNewLogicalHeight;
> +            shapeInsideInfo = currentRegion->shapeInsideInfo();
> +        }

What if regionAtBlockOffset returns nullptr?

[Zoltan Horvath]

Created attachment 219345 [details]
Patch - address Darin's comment

> What if regionAtBlockOffset returns nullptr?

It should never happen at this point. I added an ASSERT for checking it.

[Bem Jones-Bey]

Comment on attachment 219345 [details]
Patch - address Darin's comment

View in context: https://bugs.webkit.org/attachment.cgi?id=219345&action=review

> Source/WebCore/rendering/RenderBlockLineLayout.cpp:1226
> +        // flows into the next region with a shape, we need to update the current shape and region

This code seems like it belongs in the "overflowsToNextRegion" case above, as it is very strange to have two different places in the same function that move to the next region, and it also makes the code hard to follow. If it doesn't make the patch too big, I would prefer if this is refactored to make that possible. If not, then a FIXME comment here is ok, with a reference to a bug to do the refactor.

[Zoltan Horvath]

Created attachment 220463 [details]
Patch

[Bem Jones-Bey]

Comment on attachment 220463 [details]
Patch

That looks a lot better!

[WebKit Commit Bot]

Comment on attachment 220463 [details]
Patch

Clearing flags on attachment: 220463

Committed r161384: <http://trac.webkit.org/changeset/161384>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.