Root updates are next

The issue names a planned rootUpdates project and describes codecasting, but it identifies no file, test, entry point, or acceptance criteria. Start by clarifying the intended rootUpdates behavior and locating the existing Electron project context; done is not defined in the issue.

Quick update.

We went with isCarbon == false.

No doubt some scripts will behave weirdly.

We can't be perfect in porting, this is an example of a thoughtful feature -- we were trying to anticipate the future. That's where the imperfection comes from. Our rule is look up how the kernel did it, and do that. But this is a time sensitive feature, and we have no idea what the Frontier of 1998 would think about this, because it's so long gone.

Claude is very worried that there won't be backward compatibility. I have trained him well. He tests all the root files he generates against model files that came from old Frontier. It's a level of perfection I respect, even though backward compatibility at that level may not be worth that much. How many old Frontiers are still running? (I also suggested we just use the odb format with one change, the address size. Claude wanted to stay with the relational in place of hashtables. It's a good bet, I think, but we know that the hashtable approach works.

Quick update -- the table editor in Atlantis now supports basic editing. And I added a new verb, console.log, which lets me write messages to the JS console. The debugger isn't in yet.

It's really difficult managing two Frontiers, I never cleaned house on my daily Frontier for many many years (if ever). Cleaning house will make the transition more manageable. I save all the objects I don't need daily access to in a folder in Claude's docs about Frontier. And I ask Claude to get familiar with what's there. He already knows more about my code base than I remember.

This is one of the unsung uses of AI -- rebooting old code that was never economical to keep updated.

My next project -- rootUpdates. I'm using something called "codecasting," which I got working in 2007 or so. It totally deserves the name. Will report when it's working. And we're totally building on Fat Pages. Totally rational technology, really stands the test of time. Now they're enclosures on RSS feeds. See how the parts fit together. :-)