Branches with similar names missing after import


I am currently trying to import an svn repository to git and see it failing in a weird way.

When I run

subgit import "https://server/svn/repo/" repo.git

it will run the conversion until 99% and then fail.

When I take a look at the created bare git repository the repo is viable and complete except for three branches that are missing.

The original svn repository has, among many others, the following branches defined:

  • Unfinished
  • unfinished
  • Unfinished2
  • unfinished2
  • Unfinished3
  • unfinished3

If i take a look at the created git repository I see that the “unfinished”-branches with the capital-U are all missing.

Is there something weird going on with there branches being similarily named?
Is subgit doing any not-case-sensitive comparisons or something?

Hello Tim,

SubGit itself is designed to be case-sensitive and thus differentiate cases internally and does not remove any branches because of that. But if SubGit writes the data to a case-insensitive filesystem, then the file may indeed be overwritten; that may happen on Windows, for example, since NTFS is case-insensitive by default, could it be the case?
It’s not clear why the import fails at 99%, and I’m afraid that it’s hard to make any conclusions without additional information. Could you please advise what the error does SubGit show when fails? Are there any exceptions in the SubGit import logs?

Hi Ildar,

You were totally right. I spun up a linux vm and run the conversion again and this time everything was fine. You were spot on with NTFS being the problem.

Thanks for your help!

Hi Tim,

thank you for letting us know! Glad to hear that it works well now!