Apéndice F.   Una perspectiva de primera mano

Comentarios de primera mano (de personas que trabajaron en la proto-Internet) tomados de: Community Memory: Discussion List on the History of Cyberspace.

Continuando un debate anterior, el 15 de Marzo de 1999 John S. Quarterman escribe:

> Eventually there was the ARPANET-MILNET split,> and there were other military networks than MILNET.

 

This does remind me ..I got onto the net in 1986, and I remember what a chore it was trying to send email to bangpaths, the .bitnet domain, and all the other nets, via a very few email gateways whose names and methods you had to dig around to find. I remember an entire long document about how to send email from one network to another, full of long, tortuous addresses peppered with the punctuation appropriate toeach of the networks along which you intended your email to go. In this respect, DNS and the integration of the various address hierarchies was a godsend.----

 

Michael S. Hart añade:

My solution to all this was to get an account at vmd.uiuc.edu, which was BOTH a bitnet node AND an Internet node. . .too bad it wasn't ALSO a PLATO node, as I had to login on two machinesetups next to each other and swap disks to forward mail fromvmd to PLATO and vice versa. I also swapped messages to/froma BBS I was running back then. . . .That's ONE way to do "connectivity."Michael S. Hart[hart@pobox.com]

 

Project Gutenberg

Executive Director

Internet User ~#100

 

Bob Bickford añade:

Even without trying to cross network types, it could be difficult to tell someone how to email you. In 1985 my signature block had this text in it:

 

{apple,pacbell,hplabs,ucbvax}!well!rab This said to possible correspondents that my login was 'rab' on a machine named 'well' which was exactly one uucp hop away from any of the machines named 'apple', 'pacbell', 'hplabs', and 'ucbvax'. So if they had any idea how to reach those machines, then they could by extension construct a path to me. The WELL had unusually good uucp connections; many people had to give paths with six or seven bangs in them, to get from a 'known' machine to theirs. I do recall that sending to some other networks required that one know exactly what the name of the appropriate gateway was, and what oddball characters had to be part of the address and which oneswere illegal in addresses on that network, and so on. Just explaining to newusers how to construct an email 'To' line was painful. Heh -- I feel a "we had to trudge four miles in the snow" story coming on........rab-----

 

Finalmente,  Geoff Kuenning añade:

What I remember most strongly was that ihnp4 was the hub of the world. I was stuck with a uucp connection (in fact, I've still got one, though I'm phasing it out) and had to bang a lot.

 

There were two things that made it usable: (1) nearly every machine had a unique name, so you could abbreviate most paths as just destination!user, and (2) the pathalias program, "by steve bellovin,as told to peter honeyman" (lowercase in original).

 

Pathalias generated a table of bang paths to every known system in the world, so that the destination!user notation got expanded to the full sourcerouting. That worked well for uucp-only paths.

 

The moderate fun came when you had to go to a non-uucp network. But it wasn't too bad from my (non-Arpanet) end: you just stuck some magic on the tail of the uucp path, and the last site wound up with the magic and interpreted it correctly. So, for example, I could send to something like: trwvax!ihnp4!mitvax!rms@aialthough in practice there was no need to source-route an Internet message so close to its destination; instead I sent all Internet stuff through ucbvax and let it deal with the rest of the routing.

 

The *real* fun was getting to people who were non-uucp, non-Arpanet, but had an Arpanet gateway of their own. Then that simple "rms@ai" was replaced by something like "joe%watson%ibm@ibm.bitnet", which later got turned into a BITNET-style address. Fortunately, mostpeople who lived on non-Arpanet machines fairly quickly learned the magic needed by all other networks to get to their own machine, and so you'd see signature lines like:BITNET: (bitnet-style address, sorry, don't remember the syntax)ARPA: joe%watson%ibm@ibm.bitnetUUCP: ...!ihnp4!watson!joeFIDONET: etc.Many people had three alternatives, and I've seen as many as four orfive.--Geoff Kuenning  geoff@cs.hmc.edu   http://fmg-www.cs.ucla.edu/geoff/