| |
SM7UZI > BPQ 17.01.99 02:59l 305 Lines 14515 Bytes #-10140 (0) @ WW
BID : SM7UZI4617
Read: DH6BB SP6VXT DB1RAS GUEST
Subj: Re: bpq vs flexnet
Path: DB0AAB<DB0SL<DB0RGB<DB0MAK<DB0ERF<DB0SHG<DB0CEL<DB0EWB<DB0HHW<DB0IL<
OZ5BBS<OZ4PAC<OZ4BBS<OZ2BOH<OZ9BUL<OZ3BOL<OZ3BOS<OZ3PAC<OZ3BOC<SK7IQ
Sent: 990116/2228Z @:SK7IQ.MALMOE.M.SWE.EU #:62789 [Malmö] $:SM7UZI4617 FBB7.00
From: SM7UZI@SK7IQ.MALMOE.M.SWE.EU
To : BPQ@WW
>SM7UZI wrote in a message:
>You still haven't told us WHAT the reason is to use FlexNet.
It is not my duty to tell you anything!
Be happy about that you have the opportunity to get information that you
don't have the ability to collect by your self.
If you are interest reading the other bulletin I have written about what is
better with flexnet, x-net and thenetnode.
Others have also written interesting bulletins that are worth reading, both
to flex and bpq.
But I guess you really aren't, just another conservative packet user who don't
have much knowledge about packet radio and what can be done to make it better.
Who wants everything to be as it used to be, your kind will be the death of
packet radio.
>Just that you say it is good software doesn't make it better than BPQ.
>Some FACTS please.
What is better with bpq than with flexnet?
>WHAT makes FlexNet a better choice than BPQ.
Here is of some things that is better with flexnet than with bpq.
Read this interesting articles about flexnet from P.E.S.T:
http://mail.texoma.net/~n5pvl/flexdir.htm
This is an introduction to flexnet:
http://www.neosoft.com/~hezorn/
The packet driver:
It uses adaptive parameters and packet collector, which increases the
capability of link rather much.
It also makes it easy to set up, that is it build up by modules makes it even
easier. It is just to load the wanted modules in a backfire.
It uses a real time protocol called 6pack instead of kiss.
It have support for many hardware, which is good since it now a days is
meaningless to use a nc for low speed packet.
A uscc card with modems and a separate computer works as good, often even
better. At is still developed.
The node:
It uses a protocol also called flexnet that is better than the net/rom protocol
used by bpq. Not that much iver head data, flexnet uses header compression.
The routing works and is based on round trip time, not on a strange quality
measurement like in bpq that makes bpq to route wrong.
The path with shortest rtt is the fastest.
The network information is updated very often which makes the routing better.
In bpq is it updated very seldom, and the information bpq uses to decide how
route is old and, packets are not routed the fastest way. All packets is routed
the over the same link, which becomes overloaded.
The Flexnet node supports dama, can be used as dama master and slave.
There also exists a special node hardware called rmnc, which is used with
flexnet software.
It is cheap and very good for unattended operation, but also in other cases.
It is build by modules, modems cards etc. So it is suitable both for high
speed and low speed packet, you can choose processor speed etc. That is
multitasking is very good makes it good for many ports and highspeed packet.
Other hardwares capable of running as node like tnc3 are for highspeed use
only and very expensive.
>(I will not say that you make things up, but sometimes when I read your
>"arguments" I wonder...)
What exactly are lies?
I maybe also will find some lies among the bullshit you load the packet
network with.
You aren't really providing us with very much new useful information!
You should read some of the other bulletins about this subject.
>Please excuse me, but KISS has a standard.
Who have standardised it?
>Extended KISS, and other versions, is attempts to improve standard KISS,
>but I haven't seen a standard there yet.
>
>But the original KISS has a standard. And it is an easy understood standard.
Kiss is an old worthless protocol.
It's main purpose is to turn your expensive tnc in to a modem
that you could have bought much cheaper and then don't let the packet driver
used to even control it.
>To really use the standard is perhaps not that easy, as you must fully
>understand the packet-protocol you want to use.
Keep it simple stupid!
Most people prefer a good not easy understandable good protocol, than a
worthless easy to understandable one.
The reson for why kiss is so widely used isn't because it is the best available.
>Why do you think anyone are trying to get rid of KISS when they are trying
>to improve it ?
They are not!
Different host modes, hdlc , protocols like 6pack and other protocols have
been created to be used instead of kiss.
>Please tell us why you think "kiss isn't very good".
Here is some text about that from
http://www.ife.ee.ethz.ch/~sailer/ham/6pack.pdf
by Matthias Welwarsky, DG2FEF @ DB0KLN.DEU.EU,
If you are interested read the whole document.
KISS versus 6PACK
6PACK provides some advantages over KISS, most notably the ability to transmit
real time data and thus the ability to implement the channel access algorithm
in the PC.
KISS achieves data transparency by prefixing some data bytes by an escape code.
This means that the transmission capacity needed to transmit a data packet
depends largely on the contents of the packet. In the worst case, twice the
bandwith is needed. With 6PACK, the transmission capacity needed is exactly
predictable, if one neglects the real time commands inserted into the data
stream.
If one continues the comparison and compares 6PACK rings to KISS token rings,
the advantages of 6PACK become even clearer. With a KISS token ring, the
response time of the system depends on the loading of the ring. Since the token
ring does not transmit DCD and PTT states, each TNC needs to implement its own
channel access algorithm, and usually a rather primitive SlotTime/p-Persistence
algorithm is used. Also, there is no timing relation between the receipt of a
packet over the air and the arrival of the packet in the PC. The PC does not
know when the packet is actually sent. (Note by the translator: When the FRACK
timer expires and the PC retransmits a frame, the original frame may still wait
for transmission, since the channel was busy all the time. Then the
retransmission of the frame produces additional unnecessary load on an already
congested channel)
This usually leads to the transmission of frames with outdated sequence numbers
and to strange effects if one tries to implement a DAMA master on a KISS token
ring. With high HDLC bitrates and high load on the ring, the transmitter may be
interrupted between packets, which leads to a much increased collision
probability.
6PACK provides no mechanisms which would allow the TNC to implement its own
channel access. Therefore, the channel access has to be done in the PC. This
is a significant advantage, especially on half duplex links.
Short response times, which are needed by, for example, a DAMA slave, cannot be
achieved with KISS. The transmitter keying is always delayed by the time it
takes to transmit the packet on the asynchronous serial interface.
With 6PACK, real time data and channel data are distinguishable. Therefore,
the delay of real time data does not depend on the ring loading, and it can be
calculated from the ring bit rate and the number of TNCs on the ring.
Each TNC delays the byte by between 10 and 20 bits, which is 260 to 520 mys at
38.4kBit/s. The longest possible delay at 38.4kBit/s is therefore 8 * 520mys
= 4.2mys.
This means that it does not take more than 4.2mys to key the transmitter of a
particular TNC, or to communicate the DCD state to the PC. This way, even a
ring of TNCs may be controlled quite precisely, but the additional delay of
a ring may increase the collision probability slightly on a half duplex channel.
A flaw of 6PACK is that the TX delay may be longer than requested. This happens
every time a packet is delayed more than the requested TX delay compared to its
preceding "TX counter + 1 command". However, measures should be taken to prevent
the peer station from complaining about the too long TX delay.
Last, but not least: Unlike with KISS, the TNC LEDs can be controlled by the
host PC directly. They are used just like the LEDs on the RMNC3 card.
>Perhaps you doesn't know what KISS really is ?
No probably not.
>KISS is good, if you fully understand what it is.
What is better with kiss than with 6pack?
>Sure, it lacks some parameters some people wants to be able to set, but
>who hasn't a wish-list of what this or that program should be able to do.
Kiss was just a experiment that don't was meant to be used and it didn't turn
out very well either.
People used that instead of hdlc interfaces and other better protocols
>KISS was developed to make it possible to develop and test new
>packet-protocols, not just ax25. exp
Things have changed since than, other better ways to use a tnc have been
discovered.
>Host-mode is a way to give programmers an EASY way to let an application
>have access to ax25, yet making the application have the control.
>The packet-protocol, ax25, is still in the TNC.
That is how a tnc is supposed to be used.
If the packet driver is handled by the software in the tnc and it can be used
with a dumb terminal, is it used as a tnc is intended to be used.
In the former days was the tnc fast enough to handle both node software and
packet drivers.
But now a days is most people using their tncs as they where modems, the
packet driver handled by the computer.
Modern packet driver can't be handled by the tnc since it is to slow.
To buy a tnc for that use is stupid, it is cheaper and thanks to kiss does
it also work better with a modems.
And it is cheaper with a old computer and plenty of modems, than with a tnc.
>With KISS you have transferred the packet-protocol to the computer-software.
But not the full control of the modem used!
>This is often the reason why programmers thinks that KISS is difficult
>to use, as they must implement the packet-protocol completely in software.
It is since they must also make a packet driver to use it.
But to use one that already exsists and just add suport for its interface
is much more easy, at least is flexnet rather easy to use.
>The only thing a KISS-TNC does is to take care of the transport-layer,
>the rest is software in the computer.
You are using the tnc as if it was nothing but a modem, which the computer
can't control.
>This is the reason why the KISS-TNC has only a few parameters to set.
>The majority of the parameters is set in the computer-software.
>Some programmers, sadly to say, doesn't let the user set all of these
>parameters. (If they implemented the parameters at all)
To let the packet driver handle all parameters, like in the case of flexnet
is to prefer.
>This is also the reson why KISS-drivers is used as a generic packet-driver.
The reason is not that kiss is good.
To buy a tnc and then don't use it as tnc, but as a modem.
>In an other message SM7UZI wrote:
>>>>Bpq still contains bugs.
>>>Please explain the bugs.
>>Some times then you try to connect the bbs is you connected to the node for
>>some strange reason.
>What version has this bug?
the last, 4.08a.
>I have NEVER noised this bug you are referring to, and I use BPQ every day.
>>If you connect the node by its alias does it reply with it and not with its
>>call, which it should have done.
>When does this happen ?
>Have you notised these "bugs" yourself, or have someone told you that
>BPQ has these bugs ?
>Why use the alias-call in your connect ?
>Doesn't the nodes have REAL calls that you can use ?
To avoid being missrouted by the worthless routing of bpq, is there only
thing to do, to use the node alias with ssid as callsing.
C 2 alias-2, any ssid from 0-15 works.
If you use the callsign doesn't bpq care about the port specfied.
How do you manage to get around that for many major problem?
><Nit-pick-mode on>
> If you as a user, in Sweden, connects to a node by an alias you are
> violating the national amateur radio-regulations.
That is not true!
No, when calling for a station can you use whatever name you want.
But when the stations are connected must its callsigns be used,
so it must answer with a callsign.
>In the address-field of the ax25-packet it MUST be a real call, not an
>alias.
And that doesn't bpq.
> My recepy is: Never use alias in a connect-sequence.
That doesn't help much!
> You really don't need to use the alias at all, at any time.
You have to, since bpq's routing is worthless!
You often want to connect the best way, for example direct to
your neighbour and not via three other nodes.
> Just because you can do it doesn't make it legal.
> Lack of knowledge is no excuse to break any law or regulation.
> (It may be a reason WHY someone does it, but no excuse)
> Other countries may have other regulations.
That doesn't make the routing less a problem, rather the opposite!
Every one is probably aware of the rules
No but I don' think that we shall turn of every net/rom node because of that.
Use the flexnet node it doesn't use any aliases, since it is not net/rom.
I don't know any country where you don't have to answer with your call.
.:.
+-+ ### | |
\--+ +-________) (_______________________________| o|
_____/===%====/======------------------------------------!\
______________/ / //! ||||||||||||| (%) ================ ====== ! X
'______________! ! !!! ||||||||||||||| ._================_.. ====== -- ! X
\\\\\\\ o||XX|||XX||o||_! o { o } |===----------!M|
""""""\_!_!_|_!_!_!!!/ !________!-------___X____________ !-.
| "X ||__| |. ^\\ o |
73 de SM7UZI | | / |! || \\ |
Anders Lageras | /""|!!! |" || |
Lund Sweden JO65PQ "----" |!!!!! ". || |
AARS Member No. 81 |!!+-+!!| | |
Packet : SM7UZI@SK7IQ.MALMOE.M.SWE.EU |!!|:|!!| | |
Internet: anders.lageras@usa.net `--+-+--" `-"
ICQ# : 5161379 |%%#|
Homepage: http://www.visitweb.com/uzi/ |%%#|
|%%#|
|%%#|
|#%#|
|###|
|###|
+---+
o==|=====-
Read previous mail | Read next mail
| |
