Re: [TLS] TLS grammar checker?

[Peter Gutmann <[e-mail]>]

Martin Rex <[e-mail]> writes:

>To illustrate one of my many problems with ASN.1 complexity and lack of
>clarity, I was recently wondering on OCSP [...]

OCSP is a terrible spec, alongside TSP and CMP they're some of the worst
examples of ASN.1 that I know of.

OK, not quite, there are telecomms management specs that are even closer to
gibberish in terms of their ASN.1, the problem with the three above is that
not only is the ASN.1 a mess - some of the authors didn't actually understand
ASN.1 when they started work on it and just made it up as they went along -
but the text is awfully confused as well, as are some of the design features,
which were cargo-cult cut&pastes from other specs without the authors
understanding why they were doing it.  See for example my analysis of CMP in
"Plug-and-Play PKI: A PKI your Mother can Use",
http://www.cs.auckland.ac.nz/~pgut001/pubs/usenix03.pdf, for the specific case
of CMP.  I also have comments on the gibberish nature of parts TSP, posted to
the PKIX list while it was still in the draft stage (nothing was ever fixed),
http://www.imc.org/ietf-pkix/old-archive-01/msg02073.html.

(Apologies for the treasure-hunt URL refs, I'm trying to avoid quoting large
chunks of text for people who don't really care that much).

OCSP is just as bad.  So using these as examples of why ASN.1 sucks is a bit
disingenuous, you'd be hard-put to find worse examples of ASN.1 (and protocol
specs in general) than those.  If you want an example of well-designed ASN.1,
go for something like PKCS #15.

Now, can we switch to vi vs. emacs?  There's a whole lot of things there that
haven't been said yet.

Peter.