Miscellaneous questions

[Herbert Furting]

1) This is the cost of advance...
2) btw: I've never said that one mustn't provide backward compatibility.
Of course there are things that would break that (e.g. use something
else than SHA1 for fingerprints) but my ideas about how to interpret
the standard, and where to put some subpacktes wouldn't break the
formats...

On Tue, Apr 15, 2008 at 3:40 PM, Robert J. Hansen <rjh at sixdemonbag.org> wrote:
>
> > Why? Just because new (perhaps incompatible) features are added in
> > newer versions,... nobody has to use that newer versions, right?
> >
>
>  If you put GnuPG 3.0 available for download, everyone who's looking for the
> latest release will grab it.  The people who are quite happy with 1.2, 1.4
> or 2.0 won't.
>
>  Now imagine that 3.0 breaks backwards compatibility.
>
>  Anarchy ensues.  A lot of your users can't talk to each other.  Most of
> them don't know why.  "I'm using GnuPG 3.0, I don't know why it can't talk
> to PGP 5.0, I mean, GnuPG 2.0 could...!"
>
>  Ensuring a migration path is so critically important in software
> engineering.  Breaking backwards compatibility is seen as an extreme step.
>