[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: shim - transport/app communication



On Tue, 2005-03-15 at 03:01, ext Iljitsch van Beijnum wrote:

> > I believe that keeping applications shim-unaware is a worthy goal.
> 
> Some applications may want to inform the shim of their special needs. 
> But applications that don't have special needs shouldn't have to be 
> shim-aware, of course.

Completely hiding shim6 from applications might be problematic, even
though I agree that it would be a nice property. Applications should see
ULIDs, right? (Where ULID could be one of the locators). This would
change the semantics of those functions that handle addresses, such as
getsockname() or bind()? Even if using one of the IP addresses as ULID,
application gets an address that might end up to be different from the
IP address that is actually used in the communication later on. This
could break applications that have logic based on these addresses, such
as those that pass the addresses to other hosts as a part of the
application protocol message exchange.

So I'm afraid that applications that want to take advantage of shim
might need to use a different address family, at least.

- Pasi

Attachment: signature.asc
Description: This is a digitally signed message part