Re: [rdiff-backup-users] workaround for "[Errno 91] File name too long..."?

[Andrew Ferguson]

Andrew Ferguson wrote:
> address@hidden wrote:
>> On Fri, 17 Aug 2007, Dave Kempe wrote:
>>> While i agree with everything Andrew has said regarding long filenames and 
>>> such - on filesystems that don't require escaping, we have had no problems 
>>> using rdiff-backup to back up rdiff-backup repos. As long as you structure 
>>> the directorys right, its works really well.
>> 
>> When searching for the rdiff-backup-data directory, does rdiff-backup 
>> check top down, or bottom up?
> 
> rdiff-backup takes the 'dumb' approach of appending "rdiff-backup-data"
> to the second argument, not even bothering to search:

An addendum to that (and maybe this was actually your question?) is how
rdiff-backup finds the data dir during a restore:

I defer to the comment at the start of restore_set_root() in Main.py:

        """Set data dir, restore_root and index, or return None if fail

        The idea here is to keep backing up on the path until we find
        a directory that contains "rdiff-backup-data".  That is the
        mirror root.  If the path from there starts
        "rdiff-backup-data/increments*", then the index is the
        remainder minus that.  Otherwise the index is just the path
        minus the root.

        All this could fail if the increment file is pointed to in a
        funny way, using symlinks or somesuch.

        """

-- 
Andrew Ferguson - address@hidden