Re: [quicwg/base-drafts] Why is ipAddress variable length (#2158)

[ekr <[e-mail]>]

On Thu, Dec 13, 2018 at 2:39 PM Kazuho Oku <[e-mail]> wrote:

> FWIW, in #2122 <https://github.com/quicwg/base-drafts/issues/2122> we are
> discussing about using the structure to send multiple entries of
> PreferredAddress (e.g., one that contains IPv4 address and the other
> containing a IPv6 address).
>
> I'd assume that we would be doing either
>
> PreferredAddress addresses<x..y>;
>
> or
>
> struct Address {
>     enum { IPv4(4), IPv6(6), (15) } ipVersion;
>     opaque ipAddress<4..2^8-1>;
>     uint16 port;
> }
> struct PreferredAddress {
>     AddressPort addresses<x..y>;
>     ...
> }
>
> Either ways, the length of the IP address would be required to keep the
> structure extensible (i.e. allowing a new IP version to be defined).
>

Yes, I agree if you do that, then you will need some kind of length
structure. Though you could do:

```
struct Address {
    enum {...} ipVersion;
    uint8 length;
    switch (ipVersion) {


    }
    uint16 port;
}
```

This would have the advantage that it would detect malformatted addresses
directly in the parser.


-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/quicwg/base-drafts/issues/2158#issuecomment-447172636