Comments on: Subversion bulk move

By: Fred

Fred — Wed, 30 Jan 2008 12:58:26 +0000

Except for that you need quotes, that should work However, the find command will allow you to do more complex filtering than that.

By: Henrik

Henrik — Wed, 30 Jan 2008 12:34:14 +0000

So what about following small snippet?

for i in 06* ; do svn mv $i 2006/ ; done

By: Fred

Fred — Tue, 29 Jan 2008 19:27:31 +0000

Tsahi: I wasn’t using Linux.

Also, obviously using a GUI tool is always a possibility, but of course I didn’t say anything to the contrary in the article.

At last, the “find” command I was showing exemplarily here can be modified to satisfy much more complicated requirements than “filename starts with abc”, so that it makes more sense. An example would be a distinction by file creation date, file size, or a combination of them.

By: Tsahi

Tsahi — Tue, 29 Jan 2008 19:12:30 +0000

using linux looks like great fun.
with TortoiseSVN, i just drag and drop. no need to remember complicated commands.

By: Kim Sullivan

Kim Sullivan — Tue, 29 Jan 2008 13:04:37 +0000

Alex: This command actually works on the working copy, so you can still commit all the moves in one go. Unfortunately the svn mv command doesn’t accept a list of files to move, so it has to be called for each file and directory separately. GUI tools usually don’t have this limitation, but if you’re stuck with the command line client, that line is definitely a keeper.

Now if only SVN supported “true” renames, instead of simulating a file move with copy & delete

By: Fred

Fred — Tue, 29 Jan 2008 07:43:26 +0000

Yes, but that will get rid of the file history (i.e. the commit log).

By: Alex Vincent

Alex Vincent — Tue, 29 Jan 2008 02:27:48 +0000

You could also do the move on your local copy, for more than one file at a time, and then commit them all at once… can’t you?

By: fredrik

fredrik — Mon, 28 Jan 2008 23:45:59 +0000

Ah, that -exec flag is handy. I’ve found myself on numerous occasions just using find (or ls) piped through a “perl -ne” construct (or awk) to build actual Bash command strings and then pipe that through /bin/sh. Which works well, but is slightly icky. The good thing is without the last pipe to /bin/sh you can do dry runs to make sure it does what the doctor ordered.