00:17 <_whitenotifier-a> [YoWASP/nextpnr] whitequark pushed 1 commit to develop [+0/-0/±2] https://git.io/JcZjR

00:17 <_whitenotifier-a> [YoWASP/nextpnr] whitequark 9d56caa - Update dependencies.

20:51 <d1b2> <thirtythreeforty> "Philosophical" question: what's the general method for developing an SoC that mostly has the same blocks, but needs different PHYs depending on your build target? e.g. for ECP5 I want to use IDDRX2, but on Verilator I need something that just fabricates data (or a loopback or whatever). Does the top block instantiate all the PHYs depending on the platform or a flag? Or is there an nMigen idiom about how this config gets passed to

20:51 <d1b2> child modules so they can pick a PHY themselves?

20:51 <d1b2> <thirtythreeforty> (disclaimer: nMigen noob)

20:52 <whitequark> this isn't something that really has established practice on it yet

20:52 <whitequark> I think the nicest way is to have a defined PHY interface and pass that to the MAC (or equivalent) block

20:52 <whitequark> but there could be several approaches one can argue for

20:54 <d1b2> <thirtythreeforty> I will probably just take that and run with it because right now I'm in analysis paralysis, heh

20:54 <whitequark> sgtm!