Roll back or revert entire svn repository to an older revision

Check out svnadmin dump/load. It creates a text file with every version of your files. It may be possible to delete everything above/below a certain point and re-import it.

See for instance Migrating Repository Data Elsewhere

A "reverse" merge may be what you need. See "undoing changes" section of svn book.

E.g. svn merge -r 28:24 [path to svn]

If you really need to wipe 'evidence' that the files ever existed, you need to do the svndump/svnload actions described above.

In a 'normal' situation, where you made a mistake, you need to use reverse merge. This make sure that undoing the changes after r24 can also be reverted, diffed, etc.

The command below should work to undo your changes (you need to commit the result of the merge to reflect the merge in the repository)

svn merge -r 28:24

You can do a new checkout of a particular revision. http://svnbook.red-bean.com/en/1.1/re04.html

svn co path/to/my/repo -r 24

If the folder structure of your application hasn't changed, checkout the old revision and replace the .svn folders from the latest revision into the checked out old revision. Now you can commit the "older" version.

If you really want to completely remove files from the repository, you need to do an svndump into a file, filter out the revs and/or file paths you don't want, make a new repo, and svnload the filtered dump into the new repository. You'll want to carefully read the SVN book section on repository maintenance before you do any of this, and make sure you don't remove the existing repo until you're sure the new one has the stuff you want.

I hate to say this, but that is a situation where I've found myself using backups of my svn repository.

Can you copy files of a certain revision to a new directory within the repository?

here is how I would start to do it. Brutal, yes, but its the only thing guaranteed to completely ignore collisions and keep revisions history intact.

  cd /scratchdir 
  svn co -r good svn://repository
  cd /hosed_project
  svn up -r HEAD
  cat >> /tmp/cp.sh 
  ORIG=$1
  TARG=$( echo $ORIG | sed 's/\/scratchdir\///' ); 
  cp $ORIG /hosed_project/$TARG;
  ^D
  chmod u+x /tmp/cp.sh
  find /scratchdir -not -wholename "*/.svn*" -exec /tmp/cp.sh {} \;

Note, this is not the "normal" way IMO, the normal way is to create a branch from an old version, and then merge that branch back in to the head. ( at least, that's how It used to work )

Edit: the above code is untested, do NOT run it verbatim

I'm not entirely sure if this work as I haven't used it in a live production yet, but I just now tried on a test repository (I copied one of my production ones) and it seems to work.

When you're in your repository, use the following command:

svn update -r 24 trunk

Where 24 is the revision number, and trunk is the file/folder you'd like to update (or restore, in this case) to said revision number.

In my test, several files were updated and (re-)added, and after doing a commit I did not receive any warnings whatsoever. I then modified a file with some dummy text and tried yet another commit, and only said file popped up on the modified list. So it seems to work rather well!

Again, I didn't use this before in live productions, so if I'm wrong please advice. I'd love to know if this is the way to go, too, because I can see myself needing this in the (near) future.

-Dave