Fair enough - there were bugs in specs (like the X.509 letting you attach arbitrarily large image or blobs), but mostly I found that the compilers were buggy.

Fair enough - there were bugs in specs (like the X.509 letting you attach arbitrarily large image or blobs), but mostly I found that the compilers were buggy. Whenever (way back in the day) we (our/my company) did SET (secure 3P "secure" credit card protocol) with competitors (MS/HP/RSA/IBM/Netscape/etc.), because we compiled an interpreter from the spec, we were able to put in code-path switches depending on the counter party and adapt. Since they had a buggy compiler from a 3rd party - they could not. Was that an issue with ASN.1? Or crappy tooling that people used?