[lisp] Re: Paul Wouters' Discuss on draft-ietf-lisp-name-encoding-11: (with DISCUSS)

[Dino Farinacci <[e-mail]>]

Thanks for your comment Paul. See my responses inline.

> ----------------------------------------------------------------------
> DISCUSS:
> ----------------------------------------------------------------------
> 
> Thanks to Rich Salz for his SecDir review. I strongly agree with his comments.
> 
> The core problem of this document is that it specifies a namespace without
> specifying the rules of the namespace. Saying "ASCII" is not a proper specification.

The document specifies and design how LISP can encode strings. The rules are in this spec. The content of each byte are ASCII bytes.

> Is Distinguished Name (DN) the same as the X.509 meaning? I cannot tell
> from its IANA registry allocation as all that is listed there is an email
> address? :/

It is the one that that is refereneced in the document.

> If so, their format is not "ASCII", eg it is more something
> like:
> 
> "a string consisting of a sequence of attribute type/value pairs
> separated by a semicolon (';' U+003B)'.
> 
> Sometimes comma's are also considered. It also allows non-ASCII
> values. What about unprintable ASCII values, eg value 0x07 which
> is "audible bell" ? Is "ietf.name" the same as "IETF.name" ?
> 
> Why not UTF8? Or if this is deemed to have the "hierarchical properties"
> of DNS names, why not Punycode ?

I am pausing to address this since the reviewers can't agree what to reference.

> Diagram section 3 is wrongly formatted. It shows a two octet AFI field,
> followed by a two octet ASCII field, followed by a 23 bit ASCII field,
> followed by a 9 bit "0" field ? But the description and text does not
> support this.

No it doesn't it shows a "…" indicating that the ASCII string is variable length.

> Is the "." a special character ? Or "," or ";" (both used as separators
> of DNs in X.509), how about a space/tab? Or a dot (") ? Is backslash (\)
> used for masking? Is \\ supported to denote a backslash?

It means variable length and are used in many packet diagrams.

> "There are no security considerations."
> 
> What about mask-len's outside the ASCII string?

> What about mask-len pointing at the 0 octet?
> What about strings without trailing 0 octet?

Where we describe encoding, I will add text to indicate if the mask-len and null byte are not consistent and what to do.

> What about similar looking strings?

Don't know what you mean. If there are multiple encodings of DNs in the packet, the strings can be the same.

> What about privacy concerns for strings?
> What about indistinguishable Distinguished Name?
> What about a NULL name of length 1?

What about them? They are stored as they are received in the packet.

> What about an invalid length 0 that cannot include the 0 octet ?

Then the parser will jump over the AFI value. I will make that more clear.

> What about excessively long length or mask specifications?
> What about matching case sensitive or insensitive?

Well since I say ASCII encoding it is case senstive and don't need specifiy it. It would be redundant specificaiton.

> What about special ASCII characters?

What about them?

Thanks again,
Dino