> On Oct 28, 2017, at 04:00, Brian E Carpenter <brian.e.carpenter@xxxxxxxxx> wrote: > > Some of us were very badly burned, in one way or another, > by formal conformance tests of OSI implementations the > best part of 30 years ago. It can be argued that improper use of formal description techniques (FDTs) was one of the main factors that killed off OSI. http://www.cs.columbia.edu/~hgs/papers/2011/Dagstuhl%2011042.pdf http://boemund.dagstuhl.de/mat/index.en.phtml?11042 http://boemund.dagstuhl.de/mat/Files/11/11042/11042.PrasAiko.Slides.pdf In the IETF, we have evolved a few FDT usage modes that do make sense (ABNF, YANG, …), and even successfully invented some rather special purpose FDTs (RFC 4997). Research into how we can improve interoperability using FDTs without repeating the OSI mistakes should be encouraged; if something comes out of this, it also would be useful to pilot its use within IETF standardization in a way that is conscious of the pitfalls. (FDT use is highly vulnerable to what we call “process confabulation” in software engineering; constant vigilance against that is required to come up with realistic usages.) We do have some semiformal test descriptions that are oriented at interoperability, e.g. https://github.com/cabo/td-coap4 or https://www.ietf.org/mail-archive/web/6tisch/current/pdfgDMQcdCkRz.pdf These were developed after the protocol specifications were mostly done, but did help in their finishing even if they didn’t themselves flow back into the standards. We need to find *good* ways to do more of this. Grüße, Carsten