UNIX for Dummies Questions & Answers

My client is using perforce. Like CVS, perforce has the feature where you can create your own branch and check broken code into the branch with out breaking the main line build. When you finally get to a point that you are not going to break the build, you merge your branch with mainline.

Unfortunately, they are not using that branch feature here in that way. Maybe we should be but we are not.

The consequence is that I have been working on some files for a month and they are not getting backup up because they are in my perforce sandbox. They need to be checked in (for the purpose of being backed up) but cannot be because I'm not done with them yet (I don't want to break the build!).

I've never used rsync or unison before. Is there some utiltity or set of commands I can use to recursively descend my directory tree looking for files that are not set the "read only" and copy them and their directory paths to a directory that will be backed up?

It seems like the find command could do this, but I'm not sure how to recreate the directory path under some other directory. I would want to run this daily and only copy files that have changed and are read only.

thanks,
siegfried

Yes, rsync works recursively (option -r, --recursive).

If you mean filter on the permission, I don't think it is possible without first constructing an include FILE (option --include-from=FILE) piped from a find -perm 444 command.

The format of the rsync include file is not easy to understand but once you get the idea you will easily make that include file.

Good source of info about rsync:
http://www.nabble.com/Samba---rsync-f13158.html

.

As I found your question quite relevant to the use of rsync, I made a test based on my scenario above.

Not so complicated after all.

Code:

#!/bin/sh
# params
  INC_FILE="/tmp/rsync-perm"
  SRC="/home/******/scripts/test"
  DEST="/tmp/"

# construction of include file with a perm filter
  echo "+ /**/" > $INC_FILE
  find $SRC/* -perm 444 -printf "+ %f\n" >> $INC_FILE
  echo "- *" >> $INC_FILE

# rsync
  rsync --dry-run  --archive  --verbose  --include-from=$INC_FILE "$SRC" $DEST

This script will first produce an INC_FILE like this:

Code:

+ /**/
+ curl.txt
+ example
+ f1
+ fichier1
- *

One could ask why bother using rsync where a simple cp could do the job. Well, rsync is very efficient when copying across a network (check option -–compress). It checks files stats, when it finds a file that needs to be updated, it will split both files, source and target, into small chunks, checksum these blocks and will only copy the chunks that don’t match. It saves bandwidth and time. Very efficient (have a look at the speedup factor given by rsync after the process)

Many more goodies like ssh transfer and the like.

Give it a try and don’t hesitate to use the –-dry-run option, specially if you play with the numerous –-delete options! I once erased my source directory with the lethal --remove-sent-files option.