I've been thinking about it, and related to my original question on storing the comment tree in a single doc, does anyone see a problem with this approach?:
When a comment is added and a revision conflict is detected, you deserialize the latest version of the comment tree and merge the two versions programmatically, and then reattempt the update. In my case, seems this would suffice due to the comment tree size likely being small, and also not having to worry about deleted/removed comments. Thoughts? Thanks
