ShareMerge in RM9

I’m trying to get ShareMerge to work in RM9 and having difficulties. I started with a relatively small sample file and exported a GEDCOM file. I then imported this GEDCOM file to a new RM9 tree and made a couple of small edits. I then exported a GEDCOM file from the modified tree and imported it back into the original file. After that I did an AutoMerge with ShareMerge, SpourceMerge, and RepositoryMerge checked. The merge mostly worked but it had a few situations:

All of the identical duplicate individual records were properly merged between the two files
On the merge individuals, some of the events were duplicated (even ones that didn’t change)
Many of the sources were referencing the same events twice, either as copies of the citations, or another reference.

As it is, this feature isn’t very helpful. Has anyone learned any tricks to make it work better?

The only trick which might result in a small improvement is to export your master database to a new database to act as your new master having gone through the GEDCOM filter. Copy that .rmtree file to another for edits you want to test. Then import it back into the new master and see if the merge is at all improved. Compare the Enhanced Properties List of the original master to the new master along the way and after the merge. Or do so with
which should work with 9 but ignores the new Associations and Saved Queries tables.

As discussed recently in another recent thread, I discovered soon after the ShareMerge feature was introduced into RM that it wasn’t very helpful. I abandoned its use quickly after that and I have never looked back.

The only thing that ShareMerge really buys you is that it introduced a unique identifier for each person into RM so that it’s always possible to identify which people should be merged with which other people. Other than that, the problem with the feature is that it is a merge. That means that it has all the attendant merge problem of duplicate facts, duplicate sources, duplicate repositories, and duplicate source templates.

I thought when I first played with the feature that it needed to be a replace instead of a merge. That means that if you take a person out of RM to another RM database, update them there, and then bring them back into the first RM database that it should be a replacement operation rather than a merge operation. There would be no duplicate anything for the person. Even now, I can’t think of any other way the feature might work that could be helpful.