"svn cp" too smart about existing directories?

[Johan Holmberg <ho...@iar.se>]

Hi!

When I create a branch or tag I use a command like:

   $ svn cp -mMSG http://HOST/myproj/trunk \
                  http://HOST/myproj/branches/foo

But I just discovered that I can issue the same command two times 
without getting an error the second time. The *third* time I get an 
error. I had hoped that the second time would give and error.

The explanation seem to be that "svn cp" is "smart" the second time and 
detects that the directory "myproj/branches/foo" exist, and then decides 
to create "myproj/branches/foo/trunk" instead.

I have looked around and tried different ways to disable this behaviour, 
but without any success.

It seems that "svn cp" is using the same tactic as cp(1) on UN*X. But is 
this really a good idea? It seems to me that it should be possible to 
write a "svn" command in such a way that it *means* the same thing, 
independently of how what repository happens to looks like.

( I come to think about rsync(1) that has solved this by having trailing 
"/" being significant in pathnames )

Have I missed some way of accomplishing what I want?

/Johan Holmberg


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org

Re: "svn cp" too smart about existing directories?

[Ryan Schmidt <su...@ryandesign.com>]

On Aug 21, 2006, at 15:09, Johan Holmberg wrote:

> When I create a branch or tag I use a command like:
>
>   $ svn cp -mMSG http://HOST/myproj/trunk \
>                  http://HOST/myproj/branches/foo
>
> But I just discovered that I can issue the same command two times  
> without getting an error the second time. The *third* time I get an  
> error. I had hoped that the second time would give and error.
>
> The explanation seem to be that "svn cp" is "smart" the second time  
> and detects that the directory "myproj/branches/foo" exist, and  
> then decides to create "myproj/branches/foo/trunk" instead.
>
> I have looked around and tried different ways to disable this  
> behaviour, but without any success.
>
> It seems that "svn cp" is using the same tactic as cp(1) on UN*X.  
> But is this really a good idea? It seems to me that it should be  
> possible to write a "svn" command in such a way that it *means* the  
> same thing, independently of how what repository happens to looks  
> like.
>
> ( I come to think about rsync(1) that has solved this by having  
> trailing "/" being significant in pathnames )
>
> Have I missed some way of accomplishing what I want?

No, I don't think you've missed anything. Yes, Subversion does  
function the way you describe, and I believe it is indeed because  
that's the way the standard Unix cp command works. I don't know of  
any way to disable that behavior there either.

If you want to receive an error when copying to a location that  
already exists, instead of having Subversion copy into that location  
as it currently does, you could probably write a wrapper script  
around the svn command which works in this manner.


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org