Karl Palsson

To clarify, the follow patch "fixes" it, but a) this makes ethernet and wifi inconsistent and b) fairly obviously, ipv6 probably still has an issue here... (unrelated, but... obvious now)...

yes and no. devices.data file doesn't have every part ever made (yet?) You can add patterns to https://github.com/libopencm3/libopencm3/blob/master/ld/devices.data#L265-L275 locally to match your parts if you like.

Personally, I don't think it sounds like a good way to go. There's another PR that actually goes _further_ to const the descriptors, so they can be entirely in flash,...

What's actually blocking this right now? the azure build logs are gone, so i can't see what failed. I'm stuck with fedora having merged a fix for their packages, but...

When I say the "mono centos 8 repo" I actually mean this one: https://download.mono-project.com/repo/index.html from mono itself. My distro (fedora in this case) has already merged this fix, but their...

I suspect (prettttty sure) you're seeing VM overcommit effects. This can be especially obvious on smaller platforms like openwrt. With big _enough_ messages, the malloc will fail, mosquitto can report...

one correction: this repository, and the work done here all predates the ch32v003. Ie, it targets WCH's "proprietary" _two_ wire interface, similar to SWD. CH32v003 brought in _another_ proprietary interface,...

I like the general idea of this (a lot) but haven't tried this out yet.

Given that they're setting the client id specifically, that's a good bet.

(Obviously, because this relies on a patched upstream, _this_ patch can't be merged as is, the submodule SHA would be wrong, but I need to show this for reference)