Todd,

In the attached design for comment viewer, it has "all ancestor
transclusions" at the top of comment details, grouped by "item
reference". I'm not sure what algorithm to use to group this, here is
why:

Anytime someone edits a document, they clone all of the transclusions
in the current version to the new version (potentially moving them
around as they insert text, and not cloning certain ones as they
delete text). In the Deme backend, all we see is that a new
transclusion was created by the user that edited the document, on the
same item but with a higher from_item_version_number.

Thus, a given "item reference" doesn't just have a set of distinct
versions: there there arbitrarily many different <from_item,
from_item_index, from_item_version> tuples for a given to_item (i.e.
the comment we're looking at now). There is no continuity between a
transclusion that existed in version 2 of a document and the
transclusion that was cloned by the updater in version 3.Any suggestions?

Mike