On Wed, 31 May 2023 09:04:10 -0600 Jeffrey Hugo wrote: > I think Mani I looking for some "guidance" on the "architecture", and > frankly so am I. An official Ack from Jakub might not be quite the > right thing at this stage, but at-least Jakub could come in and say he > isn't planning on NACKing this right off the bat, in particular because > this functionality can be used by WWAN devices which seems to be what > caused the mess the last time around. > > We've gone full circle here. This functionality was proposed as part of > the bus. Jakub came in an NACKed that, which resulted in the WWAN > subsystem and the guidance that this functionally belongs with the > devices. I tried to put it with the AIC100/QAIC device based on that, > and that got NACKed by Daniel (GPU) saying that this belongs with the > bus. You (Greg) seemed to agree with Daniel on that. > > Fixing kernel robot tests is one thing (I haven't seen any reports on > this iteration), but if there is no agreement on where this lives, isn't > it DOA? > > In summary, if you don't like this, please give some clear guidance. > Greg, you've told me in the past that you don't discuss "architecture" > without seeing the code. Here is some code. I don't claim it is > perfect (you mentioned the QAIC version had some issues you were going > to help with), but I would like to see some input. Nothing changed here as far as I'm concerned. But while I have you -- you should probably discuss your broader engagement in the upstream community with someone like Greg. The right balance between throwing code at us and supporting maintainers. I mean real maintainers of shared subsystems, not your own stuff.
