Hello, everyone.

I am the firmware engineer and public face of a project called FujiNet, which uses an ESP32 to provide an I/O coprocessor to 8-bit systems, such as the Atari 8-bit, Commodore 64, and Coleco Adam systems.

It provides networked disks, a virtual printer, a wi-fi modem, and a network adapter with tons of protocol offloading that supports TLS and built-in XML and JSON parsing, thus allowing the target computer to enjoy a nice clean I/O channel that it can deal with at its leisure. Because the I/O interface is clean and simple, and the goal is to strive for operating system integration in every case, it can be e.g. as simple as adding some BIOS calls in CP/M.

We started on the Atari 8-bit systems, and started reaching out to other systems as we got our hands on hardware and time. The goal is to bring #FujiNet to every single 8-bit computer and game console.

#FujiNet is entirely an open source project, and while we do make limited runs of units to cover our development costs (with the only intent of breaking even, this is a labor of love), we actively work with others who can make devices for their respective communities. We do this by releasing all hardware and software on Github for public consumption.

We do not ask for licensing or any other explicit compensation. To us, having people use what is made, is enough.

A few of us have been going back and forth on trying to figure out an interface for the ESP32, which has a lot of GPIO ports, a few UARTs, SPI, and oodles of other bits of I/O, and would love to hear everyone's thoughts on a potential bus interface. The two ideas we've basically had were to use high speed shift registers with latch, or to program a CPLD to handle bus arbitration.

You can see a demo of the FujiNet, here:
https://www.youtube.com/watch?v=izdCLDQ1ZxU

The main site is here:
https://fujinet.online/

Current hardware is here:
https://github.com/FujiNetWIFI/fujinet-hardware

Current firmware is here:
https://github.com/FujiNetWIFI/fujinet-platformio

Interesting hardware. I'd like to see more about the software interface. CP/M has a CP/NET product for networking, it might be good to ensure that this product is able to service CP/NET clients and servers, as well as interface with the TCP/IP Sockets CP/NET servers and clients I've been working on.

On Thursday, January 13, 2022 at 12:30:32 PM UTC-6, Douglas Miller wrote:
> Interesting hardware. I'd like to see more about the software interface. CP/M has a CP/NET product for networking, it might be good to ensure that this product is able to service CP/NET clients and servers, as well as interface with the TCP/IP Sockets CP/NET servers and clients I've been working on.

The software interface is extremely abstracted. Each target system gets an adaptation that makes the most sense for their target system, and the protocol adapters are shared between all the targets.

There is a detailed zoom call where I spend about 3 hours walking through the code-base. :)
https://www.youtube.com/watch?v=-WhMVHUGoRE

Essentially, think of it like a mainframe I/O channel processor, highly abstracted, simple commands (OPEN, CLOSE, READ, WRITE, STATUS), and you can change the state of the channel to do functions outside of these atomics (such as accepting connections, etc.)

Douglas Miller schrieb am Donnerstag, 13. Januar 2022 um 19:30:32 UTC+1:
> Interesting hardware. I'd like to see more about the software interface. CP/M has a CP/NET product for
> networking, it might be good to ensure that this product is able to service CP/NET clients and servers,
> as well as interface with the TCP/IP Sockets CP/NET servers and clients I've been working on.

There is not much to ensure, the CP/NET protocol was kept very simple, so that even a RS-232
connection is good enough to implement CP/NET. The only restriction is that one needs to be
able to transfer 8bit data between systems. Because that is the purpose of any network technology
CP/NET can be implemented on anything. Of course one needs to understand the computer end
of whatever communication device and write the CP/NET BIOS for that.

On Thursday, January 13, 2022 at 12:45:12 PM UTC-6, thom.che...@gmail.com wrote:
> The software interface is extremely abstracted. Each target system gets an adaptation that makes the most sense for their target system, and the protocol adapters are shared between all the targets.
>
> There is a detailed zoom call where I spend about 3 hours walking through the code-base. :)
> https://www.youtube.com/watch?v=-WhMVHUGoRE
>
> Essentially, think of it like a mainframe I/O channel processor, highly abstracted, simple commands (OPEN, CLOSE, READ, WRITE, STATUS), and you can change the state of the channel to do functions outside of these atomics (such as accepting connections, etc.)

Good to know. The reason I asked was that the product information I've seen thus far indicated some things like emulation of disks and other I/O (printers, modems, etc). If ordinary TCP/IP sockets can be opened/listened, then it should suffice. This would be similar (in concept) to the WizNET module I've implemented CP/NET on before.

On Thursday, January 13, 2022 at 5:35:51 PM UTC-6, Douglas Miller wrote:
> On Thursday, January 13, 2022 at 12:45:12 PM UTC-6, thom.che...@gmail.com wrote:
> > The software interface is extremely abstracted. Each target system gets an adaptation that makes the most sense for their target system, and the protocol adapters are shared between all the targets.
> >
> > There is a detailed zoom call where I spend about 3 hours walking through the code-base. :)
> > https://www.youtube.com/watch?v=-WhMVHUGoRE
> >
> > Essentially, think of it like a mainframe I/O channel processor, highly abstracted, simple commands (OPEN, CLOSE, READ, WRITE, STATUS), and you can change the state of the channel to do functions outside of these atomics (such as accepting connections, etc.)
> Good to know. The reason I asked was that the product information I've seen thus far indicated some things like emulation of disks and other I/O (printers, modems, etc). If ordinary TCP/IP sockets can be opened/listened, then it should suffice. This would be similar (in concept) to the WizNET module I've implemented CP/NET on before.

Good Zoom meeting today, I would like to formally introduce myself, and talk about #FujiNet at the next Zoom call.
-Thom