Synching text files.

[Donovan Baarda]

On Thu, 2003-10-23 at 05:39, Kurt Roeckx wrote:
> On Wed, Oct 22, 2003 at 11:46:22AM -0700, jw schultz wrote:
> > On Wed, Oct 22, 2003 at 08:31:12PM +0200, Kurt Roeckx wrote:
> > > Hi,
> > > 
> > > I'm trying to sync text files where lines get inserted and
> > > deleted.  From what I understand, rsync uses blocks to compare
> > > and so can't find much data that matches.
> > > 
> > > Would it be possible to make it work so it finds blocks that get
> > > inserted or deleted?
> > 
> > It already does.
> 
> Does it only do it on block basis?  Or will it try to find an
> offset inside the block?

It will find matching blocks at arbitrary byte offsets.

Think of the original file as a sequence of fixed sized blocks.
Inserting or deleting a single byte "breaks" that block so it no longer
matches, but rsync will match all the blocks before and after that
non-matching block.

Note that one think that can break rsync for text files is MSDOS CR/LF
vs Unix LF line termination. This effectively makes a change every line,
and unless you have lines longer than the block size you are using,
rsync will not be able to find a single match.

-- 
Donovan Baarda <abo at minkirri.apana.org.au>
http://minkirri.apana.org.au/~abo/