[ih] .UK vs .GB

[John Day]

Actually, the layer violation in X.500 was that it tried to be DNS and Google. We knew this at the time, but between the X.500 faction that had delusions of grandeur and the PTT faction who saw it as both White Pages and Yellow Pages (and a major source of revenue), there was little chance for injecting sanity into the process.  At least we were able to get them to drop descriptive names in favor of distinguished names and we sent one person who knew something to try to help them.

Of course, the Internet screwed up DNS by making it a macro-resolver for IP addresses rather than a directory and making it something the application had to deal with. When actually it is part of the Transport Layer. It is the Transport Layer’s job to map application names to network addresses. Whether or not you want applications to be able to see addresses may be open for discussion, but they shouldn’t *have* to see them.

This is something the first Unix system on the Net (1975) got right. They hacked file_io and extended the file system, so that the syntax was “<file desc> = open(<host name>/<application-name>)”  That would have been a much better direction for the future than sockets.

Also, both X.400 and X.500 groups were none too bright.  They thought that defining the syntax of a protocol was a formal description of the protocol. (!) There was no need to define the action to be taken, it was obvious from the syntax.

I remember a meeting in the burbs of Virginia where we took them to task over this. Jim White (of SRI and the ARPANET) kept insisting that the semantics of the attributes were well-defined by the names of the attributes until we pointed out 4 or 5 different interpretations for most of the names.  I finally pointed out that according to their spec, I could put the value “Z” in every attribute and it would be perfectly fine.

No, OSI didn’t have a corner on being dense. It had its share, but so did everyone else. They had to put up with the Brits!  ;-)

Take care,
John

> On Apr 15, 2018, at 10:10, Patrik Fältström <paf at frobbit.se> wrote:
> 
> 
> 
>> On 15 Apr 2018, at 15:01, John Day <jeanjour at comcast.net> wrote:
>> 
>> The idea of knowing what type the components of an identifier are is not inherently a bad idea. It certainly gives you more information for the database design that goes with it.
> 
> Agree, but in the case of X.400 and X.500 there was from my perspective a layer violation where there was not a distinction between the email address and the name. Although the same mistake is often implemented as a convention in smtp as well when people have first.last at domain construction in their email address. :-(
> 
>   Patrik
> 
>