08:17 <Xiretza> daveshah: nextpnr-xilinx/router2 fails to route a design that has worked before (with 75e8e7f4, specifically): https://misc.xiretza.xyz/repro/dds_route_failed.tar.gz - on xilinx branch I get an assertion failure, that might be me doing something wrong though.

08:20 <daveshah> Xiretza: 8 is a lot of global clock domains. It is probably too many for the current synthesis and PnR flow. It is almost certainly the case that you should be using clock enables instead

08:22 <daveshah> I do have a long term plan to solve this, but your design is probably suboptimal in the first place (and may well result in a suboptimal placement with the vendor tools too)

08:25 <daveshah> (you might be better moving this to ##openfpga or similar where more people can give general design/VHDL advice)

08:26 <Xiretza> daveshah: hrm, alright - yeah, the clock prescaler I'm using is probably a Bad Idea™, let me fix that. If it's just a matter of too many clocks, that's fine, I was just surprised by the regression.

08:26 <daveshah> It would never have worked reliably, tbh

08:27 <daveshah> It would just be somehow going from a random placement that worked to one that didn't

08:34 <Xiretza> daveshah: works now, thanks for the quick reply!

15:20 <Xiretza> since it's a topic on ghdl gitter right now - are you afraid of legal action from Xilinx regarding X-Ray, nextpnr-xilinx, etc.? It wouldn't be the first time they go in the offensive, and previously it was intimidating enough to stop work dead in its tracks.

15:28 <daveshah> Not personally, no

15:29 <daveshah> Xray has been public for over two years, Xilinx know well about it and haven't done anything

15:40 <Xiretza> that's good to hear

16:27 <mithro> Xiretza: General advice is don't listen to engineers for legal advice. Talk to lawyers who have experience in this field and follow their advice even if it doesn't make logical sense to you.

16:38 <hackerfoo> It doesn't hurt to learn about the law a bit. The law shouldn't be opaque to the people that must follow it.

16:58 <Xiretza> mithro: I was asking less about the actual legality (deciding that is what courts are for), but more about the personal experiences/opinions/feelings of those associated with the project (who have no doubt thought about this before). If xilinx decided to sue, it'd be a bad time regardless of whether or not they're in the right.

17:14 <daveshah> I think the risk to individual developers is very low

17:15 <daveshah> In any case all my development and signing of Xilinx EULAs is done by a UK limited company with minimal assets in case it goes bankrupt

18:12 <hackerfoo> Also don't take advice from engineers about money laundering :)