A preliminary architectural idea would be:

Move the CachingWordShapeIterator nextWordEnd logic into its own CachingWordShaping segmenter, similar to ScriptRunIterator. 

Then have a new segmenter combining ScriptRunIterator and this one, and return the next word end index and the next script.

Then check canShapeByWord(script).

Agree that ScriptRunIterator needs to run first.

I'm wondering, is it possible to merge the two segmentation layers into one. I mean, currently we do:
  caching-word segmentation -> cache -> multiple segmentations -> shaping
we could do:
  script & caching-word -> cache -> other segmentations -> shaping
but can we do:
  all segmentations -> cache -> shaping
?

Is there a reason for other segmentation than script done after caching?

Performance perhaps - the segmentation before caching should be as fast as possible. After that, additional segmentation is needed to split runs into something that HarfBuzz can digest. But we can certainly experiment.

This issue has been Available for over a year. If it's no longer important or seems unlikely to be fixed, please consider closing it out. If it is important, please re-triage the issue.

Sorry for the inconvenience if the bug really should have been left as Available. If you change it back, also remove the "Hotlist-Recharge-Cold" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

This issue has been Available for over a year. If it's no longer important or seems unlikely to be fixed, please consider closing it out. If it is important, please re-triage the issue.

Sorry for the inconvenience if the bug really should have been left as Available.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

With LayoutNG approaching, I don't think we should worry about this any longer.

I'd prefer to keep this open. We can close it once we move to LayoutNG, wdyt?