11:20 <tnt> Mmm. I'm getting AssertionError: Assertion failure: port.net == nullptr (/home/tnt/projects/fpga/toolchain/nextpnr/common/design_utils.cc:94) when calling ctx.connectPort(nn.name, d.cell.name, 'D')

11:20 <tnt> but I a ctx.disconnectPort(d.cell.name, 'D') the line right before so ...

11:23 <tnt> What's weird is that this looks exactly like what 3573fcca80669c390b002c17ede911fb73c8bbbf fixed > 1 year ago.

11:52 <tnt> Err yeah, not sure how/why but disconnect doesn't work anymore.

11:53 <tnt> disconnect_port() _does_ get called but at some point, some where ... it gets reconnected.

12:01 <tnt> wtf ...

12:01 <tnt> print(d.cell.name)

12:01 <tnt> ctx.disconnectPort(d.cell.name, 'D')

12:01 <tnt> print(d.cell.name)

12:01 <tnt> the 2 prints don't print the same thing ...

12:04 <gatecat> can you provide a bit more context ?

12:07 <tnt> So this is running this thing : https://github.com/no2fpga/no2build/blob/master/sw/ice40_opt_lutdup.py

12:08 <tnt> which basically looks for LUTs output going to multiple FFs and replicate them so the LUT can be packed with the FF.

12:09 <tnt> and line 66, I added the print(d.cell.name) before and after the disconnect and ... it magically changed.

12:10 <gatecat> I think the problem is that disconnecting the port invalidates 'd', which is a reference to an entry in the net's user array

12:10 <gatecat> this is a horrible trap, I think I can fix it by passing PortRef's into Python by value

12:24 <gatecat> tnt: does https://github.com/YosysHQ/nextpnr/pull/805 fix it ?

12:42 <tnt> gatecat: yup.

12:44 <tnt> Tx :D