Hello,

The network interface could have been developed over the generic radio interface I developed to write the char driver.

And this network interface would have been generic, so write once, instanciate with all radios.

My argument was exactly that network interfaces are overkill and requires bringing in network code just to exchange a few packets.

That was not a sufficient reason apparently.

I can search it again in my archives.

Sebastien


On 18/02/2025 11:05, raiden00pl wrote:
Hi Sebastian, it's a pity that your radio framework didn't go upstream.
The lack of a standardized interface for radios is a problem now.
Network interface for small radios, especially for LORA radios is overkill.
I think Greg later admitted that because he accepted sx127x as a character
driver. There was a discussion about this topic long time ago, but I don't
remember where and how to find it now :)
Probably the best approach is to support both a network interface for
the radio and a character device. Similar to what is currently the case
with CAN bus. Both approaches have their applications and use cases.

czw., 13 lut 2025 o 09:54 Sebastien Lorquet <sebast...@lorquet.fr>
napisaƂ(a):

Hello

the hardware and driver are cool.

However, if this driver is accepted, I will be very mad, as several
years ago, I built a complete radio transceiver driver framework based
on character devices (for the si4463, using a lower/higher half
approach, and it was fully working), and this was completely refused by
telling me that I must use network interfaces.

I went as far as specifying and normalizing the various ioctl calls any
radio transceiver would ever need, everything was ready. I was also
reverse engineering the si4463 modem calculator so every setting could
be set from software, this was unique. No other driver does that.

Then, I met a brick wall.

Now these are proprietary drivers in my own private fork. I got over it.

Sebastien


On 13/02/2025 06:23, Matteo Golin wrote:
I have created an initial RN2XX3 driver here:
https://github.com/apache/nuttx/pull/15828
The design was based on some suggestions received on the mailing list
earlier in the year. I haven't added quite all of
the `ioctl` commands I would like yet, and I want to add some Kconfig
options, but I am opening the draft PR in case I
can get any pointers from the NuttX community before I get much further
along.
The implementation now as it stands works without any issues, I am able
to transmit and receive from NuttX using another
transceiver connected to my computer via USB. I even took my transmitter
for a walk and got some successful long range
test results!

Any feedback is appreciated before I add the finishing touches and
unmark as draft.
Thanks,

Reply via email to