Dean Kent (dkent@realworldtech.com) on 1/26/09 wrote:
---------------------------
>Linus Torvalds (torvalds@linux-foundation.org) on 1/26/09 wrote:
>---------------------------
>>
>>The fact is, people have done idiotic filesystems
>>like that. VMS was just one example. And it's stupid. It's
>>incredibly annoying, and it makes a simple program to
>>copy a file suddenly a hugely complicated thing.
>>
>
>Well, I don't know about VMS, but MVS has some interesting "file systems". However,
>it is my impression that most of these were implemented during a time when disk
>space was very precious, as was memory, so they were designed with that in mind.
>And when centralized computing was 'king'. Since ubiquitous networking wasn't
>an issue at the time, transmitting things around wasn't considered important either.

Hm, allow me to point you to RFC454, the FTP protocol, written in 1973.

http://www.faqs.org/rfcs/rfc454.html

It has explicit support for record-oriented and structured files. I have to say, very few real world ftp implementations supported all of those options. I had the pleasure (eh) of implementing record/stream support for the ftp client and server on our IBM mainframe back in 1986...

>So, I think you can say people learned from their mistakes - but part of it is
>that technology has marched forward much faster than those original architects likely ever considered possible.
>
>Yeah, I know Unix was designed late 60's, but the idea was to have relatively small
>machines that engineers could play on, and share files - so that seems to have been
>one of the prime motivations (rather than efficiency and cost, which is paramount
>in business, however shortsighted and foolish those decisions may seem today).

Sharing files is not a new idea. Sharing files over a network isn't that new either...