Increased code stability

[zemaj]

A key goal of Code is to stay in sync with Codex - bringing over all core improvements and functionality while adding our own.

While we've managed to achieve this so far, we've now diverged so far from Codex that keeping in sync with upstream was becoming increasingly unrealistic with the old codebase.

We've now largely completed a couple of significant refactors to;

• Separate out Code from Codex in the codebase, so we can cleaning import the parts of Codex we use while allowing us to independently expand other areas.
• Expand our test suite to cover functionality unique for Code
• Rebuild our TUI history structure from the ground up so it's faster, more reliable and portable (i.e. we can rollback, resume, undo and forward histories)

We were seeing ongoing issues with upstream syncs where we would see functional regressions as Codex made changes which conflicted with our upgrades. For example, we would often see subtle functional regressions as Codex changed systems we were using which were not easily picked up during a merge.

With the new system this should no longer be the case. We can continue to move forwards faster, taking in the parts of Codex which are useful without worrying about breaking existing functionality as we go.

[ntoxeg]

As a small suggestion, if you don't consider Code as a fork of Codex anymore, it could be a good idea to set the repo with non-fork status; forks are less findable on Github, so it would help the project visibility.

[ytwytw]

would you consider use codex as acp?
it does lose some features that i care of (history, conversation resume/tracking) but it might be able to keep codex updated with latest features