13:48 <litghost> daveshah: Just so you know, we've had the same bugs on our side, fixes are https://github.com/SymbiFlow/symbiflow-arch-defs/pull/1242/files and https://github.com/SymbiFlow/symbiflow-arch-defs/pull/1257/files

13:48 <tpb> Title: xc7: added RL_IOI_SING tile pips correct fasm_feature by acomodi · Pull Request #1242 · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)

13:48 <daveshah> litghost: ack, thanks

13:51 <litghost> daveshah: If you have a good idea on how to better fix the SING tiles, I'm all ears

13:51 <litghost> daveshah: They sort of break the consistency model we have on the other tiles

13:52 <litghost> daveshah: E.g. the top and bottom SING tiles are the same tile type, but require different bits for the same feature

13:52 <daveshah> I wonder whether rewriting the tiletypes would be an option?

13:52 <daveshah> so TOP_SING and BOTTOM_SING are considered different tile types

13:52 <litghost> Ya

13:54 <litghost> That would make the database more consistent, at the cost of specialized logic in the database generation. I'm not a huge fan of the fixup happening out in arch-defs and nextpnr

13:54 <litghost> For now, we know what the fix is, as ugly as it is

13:56 <litghost> I think maybe a post-database generation fixup would be possible. It would modify tilegrid and the tile type outputs and generate the TOP/BOT SING tiles. It would also need to emit two new segbits files, with the renames present. Not impossible, just annoying