I just completed some major revisions on a line feature, primarily due to it "Godzilla" nature (too many vertices in many records). I also removed, renamed, and added a few fields. Because of the Godzilla problem and currently awful network performance, I did the bulk of the work on my local drive. I successfully imported this local FGDB feature to the proper SDE feature dataset (using "new" at the end of its name) and reregistered the FDS as versioned. This is the first time I can remember that I did not want to rename an SDE feature after major edits because I did not change fields that are used to symbolize or label nor make SDSFIE data model changes. In other words, the maps and layers that point to it could point to the new file if I can give it the same name as the old, existing feature, saving us lots of time.
I know how to get to the old file is named ...old and the new feature just has the old name. Will this confuse/mess up the delta tables in any way? I have always assumed SDE handles versioning using some internal tracking tools, not just feature class names, but this has never come up before.
If it matters, the old file had a GUID because of some ancient reason, but it did not have a replica database. I made a text copy of the GUID contents but removed the actual GUID in the new feature.