Hi Roman, well I guess I'll have to try things out ... our current C++ stack is unfortunately quite out of sync and the people maintaining it have dropped off the face of the earth... So I'm looking for alternatives. I think the drivers themselves should be quite small, but I have no Idea if a "quite-small" PC wise is a "huge" Embedded wise.
Ideally I wouldn't want to build and sell these devices at all. Ideally some (perhaps more than one) companies would pick up the idea and build the hardware for it. The prototypes will be a little more chunky than what I hope for, but I would assume that if someone built the entire board for this use-case, it should be possible to fit it in one of these industrial sockets: https://www.distrelec.de/Web/WebShopImages/landscape_large/0-/01/Weidmuller-IE-PS-RJ45-FH-BK-P-30134280-01.jpg http://t0.gstatic.com/images?q=tbn%3AANd9GcStNfXbIDEulFrPXxR1-BKmHNMLxzJLSKXXK43xrvlLGpBKEIQPCg&usqp=CAc The general Idea would be that by bringing the protocol-adapter literally into the network socket of the unsecure device you minimize the risk of attacking the connection (Powered via PoE). My work would be funded by EU research funds and my results would therefore also be public (It wouldn't have to but I think public money should enforce public results) Chris Am 10.01.20, 03:46 schrieb "Roman Shaposhnik" <ro...@shaposhnik.org>: On Sat, Jan 4, 2020 at 2:17 AM Christofer Dutz <christofer.d...@c-ware.de> wrote: > Hi all, > > Hope you all had a good start in 2020 :-) > > This is my first posting to the project even if I’m totally a fan of it since ApacheCon NA in Vancouver … still a little bit disappointed that It’s not called MyNuke but I guess picking project names in a craft-beer-pub isn’t a great idea ;-) > > For the past 3 years I have had the joy of being able to invest all of my free time and all of my work time into initiating Apache PLC4X. > > This year I am planning on finally porting Apache PLC4X drivers into the C and C++ world. > I was planning on using Rust for this … if this is a bad idea … would be great if you could tell me. > I just want to do anything else than C and C++ natively. Rust is amazing, but this still may be an issue for a MyNewt type of a use case: https://lifthrasiir.github.io/rustlog/why-is-a-rust-executable-large.html Personally, I don't think this is a long term deal breaker, but something to be aware of for now. > My major goal is to create a MyNewt integration that allows to communicate with industrial PLCs directly. > My reasoning for this is that most PLCs are completely unprotected. So if you can plug a network cable into it, it’s actually theoretically compromised. > I would like to build little open-source protocol adapters that translate from the proprietary protocols to something secure, so we can access the PLCs in a secure way. > Because just thinking of people connecting industrial hardware to the internet in order to get data into the cloud sort of produces nightmares for me. Where do you plan to run this piece of software? What's your ideal piece of hardware? > I might even receive some EU funding to tackle this challenge. I’d love to do it with MyNewt and the ultimate thing would be > if I was able to even shrink the entire solution to something you can power with PoE and shrink-wrap into the connection cable. > > For the start I would be looking for a development board with two network connections which I can run MyNewt on (One connection can be WIFI, but I’d prefer two real ethernet sockets) > Sort of sounds stupid, but for the start money won’t be an issue … the industry is wasting millions here and if in the end the solution requires someone > wanting to use it having to invest only a few hundreds, they’d be happy about it. I think MyNewt really shines for a MCU type of a use case, once you cross into deployments where an RPi becomes an option -- well, then you've got other options too. Hence the question above. Thanks, Roman.
