I have a big old legacy GIS environment that I manage - it's been built up to its current form from at least the 10.x days, and potentially the 9.x days. For as long as I can remember, any feature classes that have an attachment relationship has looked like: FC lives in dataset, attachment relationship class and table both live at the GDB level.
Recently I've been having an issue where feature classes that are updated with new attachments fail to collaborate in our distributed collaboration due to code":500,"message":"Unable to synchronize replica.","details":["Importing delta data changes failed."]
I do have a case open with esri and we found that copy-pasting one of the FCs in question into the same dataset fixes the issue - but what happens when you do this (create a new FC with attachments enabled in our up-to-date environment) is the attachment class now appears within the dataset (with the attachment table still living at the top level of the GDB).
I am wondering if this is a documented change - entirely possible it's been this way for years and I just haven't noticed. Any insight would be great. Thanks!
