This is a follow-up of Bug 1056516. So far, there two options have been discussed in bug 1056516: (1) maintain/store extra information between neighboring hyphen-break-buffers, and access them back and forth if needed, or (2) use an AutoTArray to store hyphen-breaks information of the whole text run, so boundaries of the neighboring buffers are gone, no extra information needed, no additional overhead for accessing neighboring buffers, problem solved. Both will hurt us significantly on performance. Although extremely-long-word is not common in the real world, it would be nice to have a light-weighted implementation to solve it.

I've fixed this issue in one of the patchset in Bug 1056516, which should be landed soon. Close this.