I thought I’d provide some documentation about what happens when citations are merged improperly.
Back in January I made the mistake of running Merge Citations without doing a complete checkout afterward. The initial checks all looked fine, but you won’t really know hhow much damage has been done until you attempt to export a GEDCOM. In my case that wasn’t until a month later, so restoring to an old backup wasn’t an option.
Initially GEDCOM exports wouldn’t work at all, but some manual cleanup fixed that. Unfortunately overnight my GEDCOM export doubled in size from 60MB to 120MB. Not as bad as the first time I tried this in 2023, where GEDCOM exports went from 50MB to 989MB then aborted, but not good.
The key issue is MediaLinks. If you have 300 citations linked to 300 different people each with a media link you end up with one citation with 300 medialinks tied to 300 people, so your exported media references go from 300 to 90,000. Not a good look.
So my suggestion is until RM fixes this issue and takes medialinks into account that you not use Merge CItations with doing a backup, export a GEDCOM, merge, and then immediately do another GEDCOM export and compare the numbers. If you see exponential changes restore your backup.
Please remind me. I don’t know what “merge citations” problems were fixed in 11. The best I can remember, the only fix that has ever happened is that the automatic merge no longer merges citations with a blank citation name. But in my view, that’s a total non-fix. The only real fix would be to take media links and web tags into account when deciding which citations to merge and which not to merge. There seems to be some reluctance to adopt that approach.
But do be aware that even duplicate citations that are correctly merged without incident will become duplicated upon GEDCOM export. GEDCOM’s data model simply does not support the concept of reused citations. So if you export GEDCOM and then import it back into RM, you will have to run the merge duplicate citations tool again.
I find this situation a bit ironic in some ways. For many years, I advocated for (and perhaps even agitated for) a feature in RM to be able to reuse citations. I was frustrated about entering a citation for something like an obituary once, copying it many times for all the people mentioned in the obituary, finding a typo in my citation, and having to chase down all the copies of the citation to fix the typo in each one of them. So I was delighted to see reusable citations introduced into RM8. Little did I know how much trouble they were going to cause because the merging didn’t take into account all the relevant factors to decide whether citations were duplicates or not.
