But it also must be remembered that long before the DNS there was the hosts file and the convention in many operating systems for their hosts file uppercase only, must start wth a letter and NO hyphens (i.e. a lowest common denominator of monocase ascii) (and if the network environment was DECNET, then you only had 6 characters to play with – which in a very very large DECNET network became somewhat of a creative challenge!) I suspect that the LDH convention was the codification of compatibility with common use conventions of the hosts files, but I cannot cite a reference to this.

At the outset I’m not sure that the DNS was assumed to be 8-bit clean. I suspect that the original requirement was to support “ascii”, and the disposition of the 8th bit was unspecified.

One of (the many) IDN debates was whether it was possible to retrospectively declare the DNS 8 bit clean and stuff Unicode into the DNS payload or whether it was ‘safer” to continue with the LDH implicit 7 bit minimal common assumption and encode Unicode into this restricted character set.

As I recall there were a number of instances of DNS software that were not 8 bit clean (and years later I’ve completely forgotten which they were!).

But my suspicion was that the challenge of a suitable ascii encoding was such a neat challenge to the computer scientist in us all (!) that we accepted the challenge of devising an efficient ascii encoding without really looking at the level of resistance to 8 bit clean DNS enforcement. (All too often we make decisions about technology based on ‘legacy’ when in fact either the legacy environment is non-existant, insignificant or at end of life already!)

a host name to begin with either a letter or a digit. (to legalise 3com’s

usage of the DNS)

BTW, I have been looking for the first documentation and hopefully rationale

for the LDH restriction. All I can tell from the RFC literature is the

restriction was either a codification of existing practice and stems from

the standard host names suggested by Jon Postel in RFC 229 in response to

RFC226 from Peggy Karpin, or the restriction arises elsewhere.