On Mon, Mar 1, 2010 at 9:11 PM, Mike Baranczak <mbaranczak_at_gmail.com> wrote:
>
> I'm not minimizing the security risks. My point was that detecting the
> specific features that you need (the way it appears to be done now) is more
> reliable than detecting the library version.
> --
Ok. I think I did not write my proposal carefully enough: I agree in
that checking for existence of feature is the right thing to do, not
version.
But version known to work should be documented, as a
pre-requisite/dependency for using Jersey. Or maybe even just
recommended. But give an indication of version you are expected to
use. Sort of orthogonal issue.
Apologies for confusion,
-+ Tatu +-