FOSS Chip Design Toolset

[Ouabache Designworks]

I would like to kick off the discussion for creating a FOSS IC toolset with some assumptions that I have made about this project.

The number one assumption is that we are starting with a clean slate and designing for todays engineering environment. There is a lot of legacy crap in many of todays designs that should be deprecated. We may put in some features to support legacy code but will identify such code as "temporary" and it will disappear in future releases. If you want your IP to used in the future then you must maintain it and revise it as our tools improve. Nothing is grandfathered in.

We live in a world of design for reuse and all ip will be designed so that installing your ip can be done with an automated process.

We live in a global world where everything is exchanged over the internet. Access is controlled by supplying login credentials and digital badges. Open source ip will be read-only to the world and r/w only to the ip's owner.

IP must be packaged before it can be stored in a library or transported over the intertubes. When the moving van shows up to move you across country they do not scoop stuff off of your shelves and drop it in the van. You have to box it up first and label the boxes. Same is true for ip.

IP must be stored under revision control and backed up. That is the minimum requirement that we expect from all professionals. If you can't be bothered to do that then we don't want your ip.

Never copy data in a database. Everything is assigned to one location and anyone who needs it must access it from that location. If you copy a piece of ip then it could go viral and fragment as the design changes in some copies but not others. The only way to share ip is to clone the RCS repository. That gives you a cached copy of the design and preserves a link back to the source. Any changes must be passed up to correct the source and the source must be checked on a regular schedule for new code. Again ip must be maintained, You cannot put something in storage for 20 years and expect that it will still work when you pull it out. Bit Rot is real.

We should also come up with a project code name. Freecellera may turn into an umbrella for numerous different tools and we should have some way to keep track of all the different ones. It should also be useful for marketing. I lean toward generic SciFi names myself.

How do these align with other expectations?

John Eaton