[SVN-3912] svnadmin load does fail to process dumps with non UTF-8 path names

• Type: Bug
• Status: Open
• Priority: Critical
• Resolution: Unresolved
• Affects Version/s: 1.8.x
• Fix Version/s: unscheduled
• Component/s: svnadmin

Environment

Linux

• Assignee: Unassigned
• Reporter: Subversion Importer
• Created: 03/Jun/11 19:33
• Updated: 31/Aug/21 22:41

Description

I've got a dump from svn < 1.6 and there are revision with path names, which are
not encoded in UTF-8.
svn fails with:

svnadmin: Path '...' is not in UTF-8.

It seems to be a rare problem, but some svn clients did not correctly encode
path names in UTF-8 and svn < 1.6 did accept those revisions.
The only known workaround yet is to take a (hex)editor and find and replace
those failing revisions.
This is hard to do on "big" dumps and i don't know how many failing path names
my dump got ... may be a neverending story perhaps if there are many "incorrect"
revisions.

It would be good if svnadmin or some other tool can provide some options to do
the recoding automatically by replacing those chars with a "best" guess or some
default value (may have to be provided by the user), to be able to load such
dumps without having to edit it via editor.

http://mail-archives.apache.org/mod_mbox/subversion-users/201105.mbox/%3C1306787230.3923.4.camel@sf045%3E

Original issue reported by tkrah