Text 70, 584 rader
Skriven 2004-12-08 02:33:54 av Scott Little (3:712/848)
Ärende: FTS-5000.002 RC1
========================
**********************************************************************
FTSC FIDONET TECHNICAL STANDARDS COMMITTEE
**********************************************************************
Publication: FTS-5000
Revision: Version 2, Release Candidate 1
Title: The Distribution Nodelist
Author(s): FTSC Members, Administrator and Honoured Guests
Issue Date: 01 January 2004
Review Date: 01 January 2006
======================================================================
Status of this document
-----------------------
This document is a Fidonet Standard (FTS).
This document specifies a Fidonet standard for the Fidonet
community.
This document is released to the public domain, and may be used,
copied or modified for any purpose whatever.
Abstract
--------
Current practice for Fidonet Technology Networks (FTN) is to
maintain a nodelist used to store the details of the nodes in the
network, and the network structure.
Contents
--------
1. Introduction
2. Supercessions
3. Purpose
4. Publication and Distribution
5. Content
5.1 Comment Lines
5.2 Special Header
5.3 Data lines
6. Nodediffs
7. Segments
A. References
B. History
======================================================================
1. Introduction
---------------
Fidonet is a peer-to-peer network utilising multiple different, and
often incompatible, communication technologies that are independent
of Fidonet itself, and is organised into a particular hierarchy for
both technical and management purposes.
For direct communication, it is necessary that the originating node
know in advance the exact method and capabilities of the destination
node in order to determine if, when, and how a direct Fidonet mail
session can be established. For routing and management, the stated
hierarchy must also be known.
All this information is collected into a file called the "nodelist."
While nodelists of various flavours and completeness are in use, the
"Distribution Nodelist" is the official record of all Fidonet nodes,
their operators, their capabilities and their relation to each other
in the network hierarchy. In essence, the nodelist defines Fidonet.
2. Supersessions
----------------
FTS-0005 superseded and replaced the documents known under the names
of FSC-0002, and FTS-0002.
This document supersedes and replaces FTS-0005.
3. Purpose
----------
Along with the companion technical standard (FTS-5001) this document
defines the format and content of Fidonet's distribution nodelist.
Unlike most FTSC documents, the nodelist standards are not only
aimed at developers, but also at maintainers of Fidonet (and other
Fidonet Technology Networks) nodelist segments. While nodelist
segment maintainers should try to be quite strict in their adherence
to this document, it is recommended that software developers be
prepared to accept deviations from this standard, especially with
regard to field and line size.
It is assumed you're already familiar with Fidonet structure and
addressing terminology.
4. Publication and Distribution
-------------------------------
The Distribution Nodelist uses the filename NODELIST.nnn, where nnn
is the day-of-year of publication, starting at 001 for the first day
of January. Partial and/or alternate nodelists must use some other
base filename (i.e. in place of NODELIST).
For actual distribution, NODELIST.nnn is packed into an archive file
named NODELIST.Pnn, where nn are the last two digits of day-of-year
and P is the compression format used as listed below:
A = .arc
J = .arj
L = .lzh
R = .rar
Z = .zip
The ZIP format is presently the official Fidonet standard, although
others may be distributed in addition.
5. Content
----------
The nodelist is a flat text file containing any number of lines,
using only the ASCII (7 bit) character set.
For the remainder of this document, characters in the range 00h to
1Fh, plus 7Fh, shall be called "control characters," and characters
in the range 20h to 7Eh shall be called "printable ASCII."
Every line must be terminated with a carriage return (^M, 0Dh) and a
line feed (^J, 0Ah), in that order. The file itself must be
terminated with a single EOF character (^Z, 1Ah), and no other data
following. Future implementations should accept nodelists with the
EOF (^Z) character omitted. No other control characters are
permitted anywhere in the nodelist.
Where the operating system uses a different format for text files
software that reads the nodelist should also be able to handle
nodelists stored in that format. Where CRCs are to be calculated
(see section 5.2) the calculations must be done as if the nodelist
has the exact binary format described above.
For compatibility with certain broken segment processors, nodelist
maintainers must limit all lines to at most 157 characters (plus
terminator). It is very likely this limitation will be removed in
the future, therefore future software implementations should be able
to process lines of up to 1024 characters, and individual fields of
up to 255 characters (unless otherwise specified).
5.1. Comment Lines
------------------
Comment lines begin with a semicolon (3Bh) in the first character
position followed by zero or more alphabetic characters called
"interest flags." A program which processes the nodelist may use
comment interest flags to determine the disposition of a comment
line. The remainder of a comment line (with one exception, treated
below) is free-form ASCII text.
There are five interest flags defined as follows:
;S This comment is of particular interest to Sysops.
;U This comment is of particular interest to BBS users.
;F This comment should appear in any formatted "Fido List."
;A This comment is of general interest (shorthand for ;SUF).
;E This comment is an error message inserted by a nodelist
generating program.
; This comment may be ignored by a nodelist processor.
5.2. Special Header
-------------------
The first line of a nodelist is a special comment line containing
identification data for the particular edition of the nodelist. The
following is an example of the first line of a nodelist:
;A FidoNet Nodelist for Friday, July 3, 1987 --
Day number 184 : 15943
Please note that the above line has been split in the sole interest
of readability. It should appear on just one line.
This line contains the general interest flag, the week day, full
date, the 3-digit publication date (described in section 4), and
ends with a 5-digit decimal check value. The publication date and
check value are padded to length with leading zeros, if necessary.
The check value is derived as follows:
Beginning with the first character of the second line, a 16-bit
cyclic redundancy check (CRC) is calculated for the entire file,
including carriage return and line feed characters, but not
including the terminating EOF character. The check polynomial
used is the same one used for many file transfer protocols:
2**16 + 2**12 + 2**5 + 2**0
The CRC may be used to verify that the file has not been edited. The
importance of this will become evident in the discussion of NODEDIFF
below. CRC calculation techniques are well documented in the
literature, and will not be treated further here.
The first line will certainly be different from the above in the
case of nodelist segments for individual zones, regions, networks or
hubs; it will also be different for other Fidonet Technology
Networks.
Except for the day number and the CRC, developers shouldn't make any
assumption on the format of this line. The suggested parsing
algorithm is to find the last colon in the line, and then scan
backwards to get the day of issue.
5.3. Data lines
----------------
A nodelist data line contains no less than seven (7) variable length
fields separated by commas (2Ch), defined below. Each data line is
a record/entry for an individual Fidonet node.
Printable ASCII characters except for commas and spaces are allowed
in all string fields unless specified otherwise. Underscores (5Fh)
are always used in place of spaces (20h).
Field 1: Keyword
Type: string; Length: 1 to 6.
The keyword field defines the type of node, and will either be
empty or contain exactly one of the keywords defined below,
grouped by functional class.
No other keywords are valid in the distribution nodelist at this
time, but private nodelist distributions may include other node
types by marking them appropriately in this field. One common
example is the "Point" keyword. See FTS-5002 for the pointlist
specification.
It is recommended that future implementations allow for keywords
of up to 32 characters, and gracefully ignore any entries using
unrecognised keywords.
Full/Member Nodes -- these are the system numbers of individual
Fidonet members. See [Policy] for details on Fidonet membership.
<empty>
Defines a normal node entry.
Pvt
In the past, this keyword merely defined a Private node (see
Policy v4.07, Section 2.1.9) - i.e. a node that is
operational, but not publically available for direct contact.
Due to software limitations over Field 6, this keyword has
been "borrowed" to allow listing of systems that do not have a
PSTN/ISDN line available, but instead utilise some other form
of communication detailed elsewhere in the entry.
In either case, if the originating node cannot contact the
destination node, mail may instead be routed via its Hub or
Host.
Pvt entries are only allowed as members of local networks.
Hold
Defines a node which is operational but cannot be publically
contacted. This is expected to be a temporary condition.
Mail may be sent to such nodes, routed via its Hub, Host or
Coordinator, or held locally by the originating node until the
Hold keyword is removed.
Down
Defines a node which is not operational. This is expected to
be (semi-)permanent. Mail may NOT be sent or routed to it.
Administrative Nodes -- these nodes indicate the beginning of a
branch in the Fidonet addressing/management hierarchy.
Each branch may contain one or more administrative nodes of lower
tier, or any of the above member node types unless stated
otherwise.
Since the nodelist is a flat file, a branch is simply terminated
at the next administrative node of equal or higher tier (or EOF).
Member nodes are always part of an administrative branch, and may
never appear at the top level of the nodelist.
In order from highest to lowest:
Zone
Begins the definition of a geographic zone and defines its
coordinator. A zone is also a logical region AND net.
Normal nodes immediately below a Zone node, and before any
other administrative nodes, are Zone Independent Nodes.
Region
Begins the definition of a geographic region and defines its
coordinator. A region is also a logical net.
Normal nodes immediately below a Region node, and before any
other administrative nodes, are Region Independent Nodes.
Host
Begins the definition of a local network (net) and normally
defines its coordinator (see FTS-5001 for the override flag).
Hub
Begins the definition of a routing subunit within a multilevel
local network.
Field 2: Node number
Type: integer; Range: 1 to 32767; Length: 1 to 5.
For member nodes and Hubs (which are treated as member nodes as
far as addressing goes), this number is the node number, and must
be unique within their local network.
For a Zone entry, this number is the zone number. A region and
net number of equal value is implied, and the node number is
zero. Zone numbers must be unique across the entire network.
For a Region entry, this number is the region number. A net
number of equal value is implied, and the node number is zero.
Region numbers must be unique within each zone, but it is
recommended future Region numbers be unique across the entire
network for better 2D compatibility.
For a Host entry, this number is the net number, and the node
number is zero. As with Region numbers, net numbers must be
unique within each zone, but should be unique if possible.
Because the node number zero is reserved for Zone, Region and
Host entries, the value "0" is strictly forbidden in this field.
Field 3: Node name
Type: string.
This is the name by which the system is known.
Alternatively, this field may be used by IP nodes for a domain
name, static IP address or E-Mail address for email tunnelling
programs.
NOTE: There may be formatting limitations on this field for IP
capable systems; consult the section on IP flags in FTS-5001.
Field 4: Location
Type: string.
This is the physical location of the node.
Generally, this is expressed as the primary local location (town,
suburb, city, etc.) and where applicable, an underscore followed
by the abbreviation of the regional geopolitical administrative
district (state, province, department, county, etc.).
Field 5: Sysop name
Type: string; Length: 1 to 36.
This is the name of the Fidonet member responsible for the node.
Field 6: Phone number (PSTN or ISDN)
Type: string; Length: 3 to 29 (theoretically); Characters:
digits and dashes, or the exact string "-Unpublished-"
This field contains two or more numeric subfields separated by
dashes (2Dh). The first field is the country code. The rest of
the phone number should be formatted according to local customs.
The various parts of the phone number are frequently used to
derive cost and routing information, as well as what number is to
be dialled. A typical example of the data in a phone number field
is 1-800-555-0100, corresponding to country 1 (USA), area 800
(area code), exchange 555, and number 0100.
Alternatively, this field may contain the string "-Unpublished-"
if the node has no conventional phone number, and uses the "Pvt"
or "Hold" keywords. It is recommended that software be able to
recognise this string as an indicator to never attempt to dial
this node by PSTN/ISDN, no matter what the keyword.
This field may also contain the static IP address of an IP-only
node in decimal format with periods replaced with dashes, and
prefixed with a country code of 000. This method should only be
used in exceptional circumstances, since it risks conflict with
conventional dialling, and is almost always better to list a
fully qualified domain name in a field that allows it.
Field 7: DCE speed (required, but obsolete)
Type: string; Length: 1+
In the past, this field was used to show the maximum modem speed
supported by the node, but has since been obsoleted in favour of
the more accurate modem flags.
Commonly accepted values in this field are: 300, 1200, 2400,
4800, 9600, 14400, 16800, 19200, 28800 and 33600. These values
are normally enforced by segment processing software; deviations
risk the entire entry being declared in error and dropped from
the nodelist or segment.
Systems without a listed PSTN line must use 300, including ISDN
and IP only nodes. All others may use whichever value is most
appropriate. Because this field is now of little value except
for human readers, most PSTN capable systems currently list at
most 9600. Nodelist maintainers should confirm with their
Coordinator(s) before listing higher rates.
Field 8+: Flags (optional)
Type: string; Length: varied; Characters: varied.
Any remaining fields from position eight (8) onward are flag
fields. Note there may or may not be a trailing comma after
Field 7 if there are no flags listed.
See FTS-5001 for details on the flag fields, including length
restrictions.
6. Nodediffs
------------
The nodelist, even in archive form, is a substantial document (or
file). To reduce bandwidth usage/waste and transmission times, a
smaller file called a "nodediff" is generated using the previous
week's nodelist. The nodediff is actually an editing script which
contains only new or changed lines between the two nodelists, plus a
number of editing commands.
The nodediff file is named and archived in the same way as the full
Distribution Nodelist, except with a base filename of NODEDIFF.
The first line of NODEDIFF.nnn is an exact copy of the first line of
LAST WEEK'S nodelist. This is used as a first-level confidence check
to insure that the right file is being edited. The second and sub-
sequent lines are editing commands and editing data. There is no
terminating EOF (^Z) character on a nodediff.
There are three editing commands and all have the same format:
<command><number>
<command> is a 1-letter command; A, C, or D. <number> is an
integer from 1 to 32767 (inclusive), and defines the number of
lines to be operated on by the command. Each command appears on a
line by itself. The commands have the following meanings:
Ann - Add the following nn lines to the output file.
Cnn - Copy nn unchanged lines from the input to the output file.
Dnn - Delete nn lines from the input file.
The following illustrate how the first few lines of NODEDIFF.213
might look:
;A Friday, July 25, 1986 -- Day number 206 : 27712
D2
A2
;A Friday, August 1, 1986 -- Day number 213 : 05060
;A
C5
This fragment illustrates all three editing commands. The first line
is the first line from NODELIST.206. The next line says "delete the
first two lines" from NODELIST.206. These are the identification
line and the line following it. The next command says "add the next
two lines" to NODELIST.213. The two data lines are followed by a
command which says "copy five unchanged lines" from NODELIST.206 to
NODELIST.213. Notice that the first line added will ALWAYS contain
the new nodelist's CRC.
Since only the differences will be distributed, it is important to
insure the accuracy of the newly created nodelist. This is the
function of the CRC mentioned above. It is sufficient for a program
designed to perform the above edits to pick the CRC value from the
first line added to the output file, then compute the CRC of the
rest of the output file. If the two CRCs do not agree, one of the
input files has been corrupted. If they do agree, the probability is
very high (but not 100%) that the output file is accurate.
7. Segments
-----------
Segments are partial nodelists, usually containing only one complete
branch of nodes, starting at the relevant zone, region, host or hub
node. The Distribution Nodelist is also referred to as a composite
nodelist, as it is compiled from multiple zone-lists.
The format and structure is the same as the Distribution Nodelist,
including the header, with a base filename of local invention,
usually derived from the contents of the segment (e.g. N712LIST.nnn
or N712SEG.nnn for a segment containing all the nodes in Net 712).
To generate the Distribution Nodelist, segments propagate up through
the network's administrative hierarchy, each tier compiling segments
from the tier immediately below, then passing on the resulting
segment to the next tier upstream.
For example, a Network Coordinator will collect Hub Segments from
all the hubs in the same local network, compile it into a Network
Segment, and forward it on to the Regional Coordinator. And so on.
Segments can also be sent back downstream - this is most commonly
done with Zone Nodelists. Nodes that do not need to attempt direct
contact with nodes in other zones can use the Zone Nodelist instead
of the full composite nodelist, saving on storage space and
transmission time.
A. References
-------------
[FTS-0005] "The distribution nodelist", Ben Baker, Rick Moore.
February 1989. Obsoleted by this document.
[Policy] "FidoNet Policy Document" v4.07 - June 9, 1989.
B. History
----------
Rev.1, 19990627: Initial Release. Principal Author David Hallford
Rev.2, 20000424: Re-draft by Gino Lucrezi, with input from others,
especially Andreas Klein. Major changes:
- notes on parsing line 1
- baud rate
- different use of fields for IP nodes
- notes to developers and maintainers
- notes on pointlists
- notes on line and field limits
- revised definition of "Hold" nodes.
- clarification on hub node numbers
- clarification on point phone numbers
- clarification on the former "speed" field
Rev.2, 20040817: Re-re-draft by FTSC.
- Added Introduction and Segments sections.
- More allowances for IP listings.
- Reorganised/reindexed Content section.
- Added length limits to fields where they could
be determined.
- clarification on usage of Keywords.
- clarification on special node numbers.
- ZCs authorise that ZIP is now the default
archival/compression tool for distributed files.
**********************************************************************
-- Scott Little [fidonet#3:712/848 / sysgod@sysgod.org]
--- GoldED+/W32 1.1.5-31012
* Origin: Cyberia: 100% Grade "A" mansteak baby. (3:712/848)
|