Nathan Fiedler <nfiedler@bluemarsh.com> writes:
> I created the respository again and tried the same import again. Still
> fails with the same error message:
>
> >>>>>>>>>>>
> subversion/libsvn_ra_local/ra_plugin.c:145
> svn_error: #21093 : <Couldn't find a repository.>
> Unable to open an ra_local session to URL
>
> subversion/libsvn_ra_local/split_url.c:92
> svn_error: #21093 : <Couldn't find a repository.>
> svn_ra_local__split_URL: Unable to find valid repository
> >>>>>>>>>>>
Somebody else reported this earlier today. I can't figure it
out... he's able to 'svnadmin create foo' and 'svnadmin lsrevs foo',
so berkeleyDB isn't having problems opening the repository. But
'import' is having problems for him still....?
The funny thing is, if you can run the python tests successfully,
that's an example of 'svn import' working correctly over ra_local.
The python tests always start out doing an import to a file:/// url.
Here's what I suggest: try gdb'ing the 'svn import file:///'
command. Set a breakpoint on split_url.c:74. (If you built your
client shared, you need to set a breakpoint on main() first, *then*
set the real breakpoint.)
The loop at split_url.c:74 is simple to understand: it tries to call
svn_repos_open (which calls svn_fs_open_berkeley) on a particular
path. If the attempt fails, it strips a component off the end of the
path, and attempts to open the parent. It keeps doing this until it
runs out of components to strip, or until it successfully opens a
repository.
What you want to do is look at the value of *err after each call to
svn_repos_open, and see what the error message is. Hopefully, one of
them will be a detailed Berkeley error of some kind... that's the
magic info we're looking for.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Apr 9 22:48:45 2002