From Leonard Erickson · 1996-04-12 11:00 · #73181 in FREE25
In reply to How a new nodelist would work..... by Jason Steck
-=> Quoting Jason Steck to Pierre Phaneuf <=-
JS> Both of the above comments indicate that you're not grasping the full
JS> extent of what I am talking about. I am talking about making it
JS> possible for people to use the Internet as a transport even if they
JS> aren't full-time Internet sites. For example, I am not a full-time
JS> Internet site and I don't have an IP address. But there are a variety
JS> of non-Telnet ways to use the Internet as a transport for mail to me
JS> (theoretically -- not all of these will work as of this moment, but
JS> we're talking about the future here):
JS> 1) You could MIME or UUENCODE a mail packet into an e-mail to
JS> "tutakai@oasis.novia.net" 2) You could FTP it to an inbound directory
JS> that I could specify on my ISP. (anonymous login -- good for unsecured
JS> inbound mail). 3) You could send it in any RFC-822 format through a
JS> UFGate-type arrangement in my local area.
JS> etc. The point is that more than just an IP might be necessary. The
JS> e-mail address in the "mapped" section of my nodelist entry would
JS> specify that I was able to handle MIME/UUENCODE packets embedded in an
JS> RFC-822 e-mail. An FTP address (complete with path) would specify an
JS> FTP-accessible inbound area.
I think that the sorts of tricks you refer to above should *not* be
listed under any "physical address" section. Instead, I'd have a "n/a"
entry in the physical address table, and point the appropriate logical
addresses at *that*.
I see something like this:
Physical address(es):
Telephone
Telex/TWX
IP address
others...
n/a
The physical address info would be a type identifier followed by the
address. While text takes up more room in some ways, it provides enough
extra flexibility that I think it's preferrable to binary.
Logical address(es) info.
The logical address, a network ID, and a pointer to the
appropriate physical address.
protocol info (message/data formats also)
protocol ID, other related info, pointer to logical address
that supports the protocol. Logical address because some protocols and
formats require embedded logical addresses.
So, for your examples, the physical address would be n/a. The logical
address would be a domain-style Internet address (or even a bang path
address!) pointing at the physical address. And the protocol info would
be MIME, UUENCODE, and FTPMail pointing at the logical address. The
FTPmail entry would include a host address and directory as part of the
"related info".
You might want to take a look at the format of UUCP map entries to get
an idea of how this sort of thing works.
BTW, note that my format has the "advantage" that you don't need any
particular order to the items. For example, you can throw out all the
protocol entries that refer to formats you can't use. You can do the
same with physical addresses. And with logical addresses pointing to
them.
... Docs? Last time I went, I got nasty tasting medicine.
--- Blue Wave/DOS v2.30
* Origin: Shadowshack (1:105/51)