Network Protocol?

[DaveC]

I know this is the wrong group but I need a little pointing in a direction
please.

I need to write an implementation of a yet to be published protocol. It is
transported over the internet via IP. Can any one help me with ideas of how
to write such drivers in Windows? or if that's not doable should I develop
it in Linux? This is just for testing.

Thanks
DaveC

 

[John Harrison]

I know this is the wrong group but I need a little pointing in a direction
please.

I need to write an implementation of a yet to be published protocol. It is
transported over the internet via IP. Can any one help me with ideas of how
to write such drivers in Windows? or if that's not doable should I develop
it in Linux? This is just for testing.

Thanks
DaveC

One of the groups in would be the place
to ask about writing Windows drivers.

john

 

[Phlip]

I know this is the wrong group but I need a little pointing in a direction
please.
http://groups.google.com

I need to write an implementation of a yet to be published protocol. It is
transported over the internet via IP. Can any one help me with ideas of how
to write such drivers in Windows? or if that's not doable should I develop
it in Linux? This is just for testing.

I have a very long shotgun, but when I put the barrel in my mouth I can't
reach the trigger. Should I attempt this under Windows or Linux? This is
just for testing.

Why the hell do you need a new wire-level protocol? Just stuff your data
into XML and mail it out on the HTTP protocol. I could write a server for
that in my sleep.

 

[marbac]

Phlip wrote:

Why the hell do you need a new wire-level protocol? Just stuff your data
into XML and mail it out on the HTTP protocol. I could write a server for
that in my sleep.

Let me inform you, that the ip layer is not a "wire-level-protocol". The
"wire-level" is layer 1 (physical layer)and maybe some parts of layer
2 . tcp/ip is arranged in layer 3 and 4 (Network and Transport layer).

regards marbac

 

[Phlip]

Phlip wrote:

Let me inform you, that the ip layer is not a "wire-level-protocol". The
"wire-level" is layer 1 (physical layer)and maybe some parts of layer
2 . tcp/ip is arranged in layer 3 and 4 (Network and Transport layer).

Whoops; you're right. My mistake.

The OP should _definitely_ invent a new protocol at the network and
transport layer.

 

[Jerry Coffin]

I know this is the wrong group but I need a little pointing in a direction
please.

I need to write an implementation of a yet to be published protocol. It is
transported over the internet via IP. Can any one help me with ideas of how
to write such drivers in Windows? or if that's not doable should I develop
it in Linux? This is just for testing.

In a suitable newsgroup (e.g. comp.os.ms-windows.programmer.win32) ask
about the Winsock SPI (Service provider Interface).

 

[The Wizard]

I need to write an implementation of a yet to be published protocol. It is

transported over the internet via IP. Can any one help me with ideas of how
to write such drivers in Windows? or if that's not doable should I develop
it in Linux? This is just for testing.

Is it an application layer protocol ? A middleware ? A transport protocol ?

Because if you wanted to use TCP as a transport protocol you could take
advantage of the well known socket API that you can find in every Unix like
distribution...

The Wizard

 

[DaveC]

Whoops; you're right. My mistake.

The OP should _definitely_ invent a new protocol at the network and
transport layer.

Yes I should .. you should stop asking questions and go back to those Alt
news groups.

DaveC

 

[DaveC]

Is it an application layer protocol ? A middleware ? A transport
protocol ?

Because if you wanted to use TCP as a transport protocol you could
take advantage of the well known socket API that you can find in every
Unix like distribution...

The Wizard

I said IP and I meant it, I was not joking like many people here seem to
think? I did'nt expect such a hostile responce for asking a question.

DaveC

 

[DaveC]

(e-mail address removed) (Jerry Coffin) wrote in

In a suitable newsgroup (e.g. comp.os.ms-windows.programmer.win32) ask
about the Winsock SPI (Service provider Interface).

No I will not ask about sockets, AKA TCP/IP.

DaveC

 

[Phlip]

Yes I should .. you should stop asking questions and go back to those Alt
news groups.

Since '97 I have helped folks, on this newsgroup, provide business value to
their stakeholders.

The value of creating a new protocol at the IP level is negligible. You are
discussing adding another control layer; a peer of TCP and UDP. The cost is
extraordinarily high, beginning with the cost of persnickety driver
development.

Now, what business value do you expect this new protocol to provide?

 

[Joe Laughlin]

I said IP and I meant it, I was not joking like many
people here seem to think? I did'nt expect such a hostile
responce for asking a question.

DaveC

I'm confused about why you need to write a "driver" (by which I mean a
device driver) to implement something that's running over IP.

Anyways, this is a C++ newsgroup, discussing standard C++. Your question is
off-topic.

 

[DaveC]

Since '97 I have helped folks, on this newsgroup, provide business
value to their stakeholders.

The value of creating a new protocol at the IP level is negligible.
You are discussing adding another control layer; a peer of TCP and
UDP. The cost is extraordinarily high, beginning with the cost of
persnickety driver development.

Now, what business value do you expect this new protocol to provide?

I'm sure you are very good at what you do, I'm just glad you were not
around to count dollars and cents when TCP was invented. Where would we be
now if that were the case?

Further, I typicaly read something before commenting on its value. I hope
I'm not the only person left who places science above corporate interests
or is that a false dream in todays world?

But if you insist that I talk in your "lingo". The protocol I am writing
will plug a wide open hole for a particular sector of the market place. If
you looked hard enough you would see it also. Prior to this no agreeable
standard has been found so I hope to propose one.

DaveC

 

[Phlip]

But if you insist that I talk in your "lingo". The protocol I am writing
will plug a wide open hole for a particular sector of the market place. If
you looked hard enough you would see it also. Prior to this no agreeable
standard has been found so I hope to propose one.

Vetting science against business interests is fun. (BTW TCP sucks, or we
wouldn't have trunks crammed with spam, malware, and attacks these days.)

I would do this by finding an IP simulator that works in the user space,
above the driver level. Then I'd write the XCP protocol or whatever, and
integration-test it against concurrent packets of TCP and other existing
protocols. Only after w

You won't admit what it does or what it's for, right? Note that TCP/IP was
invented in the early 70s. Hardware is a little faster these days. IPv6
might already have what you need, and a simple transport protocol on top of
the sockets layer might also already have what you need.

 

[Phlip]

Vetting science against business interests is fun. (BTW TCP sucks, or we
wouldn't have trunks crammed with spam, malware, and attacks these days.)

I would do this by finding an IP simulator that works in the user space,
above the driver level. Then I'd write the XCP protocol or whatever, and
integration-test it against concurrent packets of TCP and other existing
protocols. Only after

....fully excersizing the new protocol would I commit to implementing it
inside a real driver layer.