Content pfp
Content
@
https://opensea.io/collection/dev-21
0 reply
0 recast
2 reactions

July pfp
July
@july
Coming from a robotics / embedded side of things - ofc i started thinking about how to do this, beyond just a prototype, my mind wanders: - well, at some point the FADEC is going to have to be on a microcontroller, should I overoptimize today, and write it all in C right now, and write the PID controller etc in C? yeah i've done that before - should i use RTOS like Zephyr? that'd be fun - but then I need the microcontroller to talk to each other, oh right what am i going to use? - how about protobuf? zmq? ROS2? (absolutely out of the question) i thought about Zenoh as well, but then I'd have to sue zenoh-pico, annoying. - what's that xkcd meme? 14 protocols, let's build a new protocol - also protobuf the data packets that aren't sent don't have a set size upfront, so maybe zmq + nanopb, nanopb is flexible and they have fixed array sizes so you can allocate memory upfront (another reason embedded folks don't use dynamic memory allocation upfront C++ 'new' as well as C malloc. no wonder everyone hates C++ in embedde world. just allocate the memory upfront - why not rust? you love rust July - yeah i know i know, but here's the thing, RTOS rust limited board targets today - why not RISC-V july? yes yes I know things are improving, but a lot of my existing experience is in arm64, cortex m0~m3 environments and i'm used to it, shoot me what packet routing will we use? how much bandwidth will it take? - and this is just the avionics side, not even doing the mechanical yet...
2 replies
2 recasts
18 reactions

July pfp
July
@july
ironically, one of my biggest reasons i won't use RISC-V today, is they just started supporting arm64 github action runners, and yes I can setup my own software-in-the-loop and hardware-in-the-loop tests, this means in the future i'll be easily be able to setup arm64 release servers, that create the binaries that I can use to OTA update my microcontrollers, and frankly, i'm actually excited about that. it means more testing, and less of i have to build my own RISC-V release servers / testing. choosing technology that's just becoming a standard vs bleeding edge stuff. another reason
2 replies
0 recast
6 reactions

July pfp
July
@july
this is absolutely trivial reason, but it illustrates something really important to me. if the technology isn't widely supported, i'm going to have further annoyances later on. which means that i'll have to build it myself, and if it's not a core competency or directly serves our core value proposition it will take away time effort, people's time and resources into things that essentially don't matter, and that's why i wouldn't do risc-v for something like this today
0 reply
0 recast
2 reactions

Lee pfp
Lee
@neverlee
my brain missed the memo, but my heart gets it.
0 reply
0 recast
1 reaction