> On 2004-03-09 12:10:41 +0100, Erik Huelsmann wrote:
> > The axiom should probably read 'disk space is cheaper than
> > bandwidth' which holds even on PDA's which presumably are going to
> > use mobile phone connections to synchronize with repositories.
> > Retrieving or sending full text-bases will make the Subversion
> > communication protocols *much* slower than they currently are.
> > Removing the text-bases will remove the ability to send diffs over
> > the wire...
> This is wrong. If the file has not been modified, nothing needs to be
> transfered for this file. Moreover one doesn't have diffs for binary
> files (the text-base would be useful for "svn revert", though). The
> point is that one can save disk space for things that wouldn't require
> bandwidth in practice (most of the time).
Why would I be wrong? Subversion *does* do diffing for binary files. Even
stronger: Subversion does not know the difference between text files and binary
files (except for it's diff and merge operations). I never said it uses
'diff' to calculate the differences. It determines the difference between two
files - treating them as binary - and sends that. This is not something you only
can do with textfiles.
The text-base is usefull for non-networked operation with diff and revert as
well as determining which files actually changed: How do you propose to tell
whether a file has been modified if you have nothing to compare it with?
As an aside: CVS sends the contents of all files it *thinks* are modified
over the wire to determine if that is really the case. Subversion sends the
difference in the files it *knows* are modified. There are people who noted that
Subversion is much more useable in a mobile way because of that.
+++ NEU bei GMX und erstmalig in Deutschland: TÜV-geprüfter Virenschutz +++
100% Virenerkennung nach Wildlist. Infos: http://www.gmx.net/virenschutz
To unsubscribe, e-mail: firstname.lastname@example.org
For additional commands, e-mail: email@example.com
Received on Tue Mar 9 13:22:51 2004