On 19/01/07, Mark Baker <[EMAIL PROTECTED]> wrote:
>
>
>
>
>
>
> On 1/19/07, Steve Jones <[EMAIL PROTECTED]> wrote:
>  > On 19/01/07, Mark Baker <[EMAIL PROTECTED]> wrote:
>  > >
>  > >
>  > >
>  > >
>  > >
>  > >
>  > > On 1/19/07, Mark Baker <[EMAIL PROTECTED]> wrote:
>  > >  > On 1/19/07, Gregg Wonderly <[EMAIL PROTECTED]> wrote:
>  > >  > > It is an enabler of content
>  > >  > > transfer, which in some case might be mobile code, but the fact 
> that any
>  > >  > > particular document type represents code is something that the 
> client
>  > >  > > determines, not REST.
>  > >  >
>  > >  > Actually, it's something the server (and therefore, message, and
>  > >  > therefore the connector, and therefore REST) declares, e.g.
>  > >
>  > >  oops, s/connector/data
>  >
>  > I'd be a bit worried from a security perspective if it was the server
>  > that purely determined remote code execution,
>
>  The server just says "this is code", it doesn't ask the client to
>  execute it, nor would it care if the client did or not, since that's
>  not part of its responsibilities as a server.  I think that's
>  fundamental to all client/server architectures, including REST.

But surely then this means that "mobile code" in REST is just "special
data type", which IIRC is different from the JERI/Jini approach where
mobile code is a fundamental part of the architecture.  If all "mobile
code" consists of is shift data with a marker saying "this is code"
then pretty much every data shifting element ever conceived could
claim to support mobile code.

Reply via email to