Björn Kautler

The output says `Found 3077 commit-pointing refs`, `Updating 3075 Refs` and `Updating references: 100% (3075/3075)`, so it might actually "just" be two refs that were not updated. Who knows though...

Ah, this seems to be caused by different refs with same name but different capitalisation it seems.

It still says 3075 instead of 3077 rewritten refs, but I guess one of them is `HEAD`, no idea what the other is. But the wrong entries in the `object-id-map.old-new.txt`...

Here you have a fully self-contained example in one line, just adapt the path to the JAR: ```bash mkdir foo && cd foo && git init && touch a b...

Well, because that is not the purpose of the file. The purpose of the file is a mapping of "old commit" to "new commit". For `empty1` and `emtpy2` there are...

Is https://github.com/jwiegley/git-scripts/blob/master/git-remove-empty-commits any better or faster than just doing `git filter-branch --prune-empty --tag-name-filter cat -- --all`? If not, then yes, due to speed. Using format-branch is awefully slow when BFG...

Or if you don't want to change that file, how about a second file that lists the pruned commits one per line, that would also be sufficient.

It needs more than an hour just to do the command from your second link and then I'd never know if this is really 100% the same BFG will do...

Ok, I learned enough Scala and trusted my beloved IntelliJ to even come up with a patch now. :-) [make_pruned_commits_differentiatable_in_report_file.patch.gz](https://github.com/rtyley/bfg-repo-cleaner/files/2159624/make_pruned_commits_differentiatable_in_report_file.patch.gz) ```patch diff --git a/bfg-library/src/main/scala/com/madgag/git/bfg/cleaner/ObjectIdCleaner.scala b/bfg-library/src/main/scala/com/madgag/git/bfg/cleaner/ObjectIdCleaner.scala index 9e61007..d6f544e 100644 --- a/bfg-library/src/main/scala/com/madgag/git/bfg/cleaner/ObjectIdCleaner.scala...

Sure @javabrett, I made a PR to your PR branch :-)