docs(rehab-log): correct two factual claims - #1
Conversation
- The archive-triage line said 'months later, nobody has needed anything back' — it has been days, and agent-era feature work off the same lineage (the ITER-0005 persistence arc) was since grafted onto main and underpins active design. Corrected to reflect that the scaffolding was disposable while the code beneath it was not. - 'It never had a reviewer' understated the process layer (PAR adversarial audits, iteration log, behavior-scenario corpus); corrected to note the review/roadmap lived inside that layer, illegible only to a reader who never opened it. Keeps Matt's structure and voice; fixes the record on our fork.
There was a problem hiding this comment.
Thanks for the corrections, and I fully support it. I would like to take a moment to coordinate since we both made corrections concurrently.
I'm incorporating your PR#1 changes into my forked version. I'll share the full changes soon so you can get eyes on it. I think it's a much clearer characterization of the work you did, but let me know your thoughts.
|
@nnunley as promised, below is the adjusted and refined version. Personally, I think this prose and framing is a more honest fit than the original. Let me know. . Published log (on my fork so you can easily view it; PR to house it in your repo can easily be submitted; OK if you prefer it exists only on fork- let me know) https://mparrett.github.io/fmpl/fmpl-rehab-log.html And not to clutter up this PR's context, but related- here's a draft (not thoroughly reviewed) focused on my dev process navigating the agentic bits that we weren't able (or willing) to spin up. I think it does a better job highlighting the excellent harness/environment work you did to guide the earlier development. Let me know your thoughts/preferences- your baby! https://mparrett.github.io/fmpl/fmpl-process.html see mparrett#9 for more details on the changes (intended advisory only). |
|
Heads up that #2 also touches |
|
Merged — thanks for the corrections, both are better than what they replaced. Two notes on mechanics, since I touched the branch: it was four commits behind Took action on this one since it was straightforward, and it reduces the handful of outstanding artifacts I've already thrown at you — but let me know if you'd have preferred to merge it yourself. |
2 participants
@mparrett — your rehab field log is genuinely excellent engineering writing, and the rehab behind it (build hygiene, CI, the gap ledger, the determinism fix, "sideline don't delete") was exactly right. I'd like to correct two factual claims that couldn't have the benefit of the project's fuller history. Both edits keep your structure and voice — only the two sentences change. Please review, tweak, or push back; I opened this as a PR rather than editing your words silently.
§2 — the archive-triage line
It read "months later, nobody has needed anything back from the archive." It's been days, not months — and agent-era feature work off the same lineage (the ITER-0005 persistence arc: durable TupleSpace, content-addressed recovery, bytecode persistence, on branches this fork predated) has since been grafted onto
mainand now underpins active design. The correction distinguishes the disposable scaffolding from the code beneath it.§7 — the "never had a reviewer" line
The agentic process actually carried its own review and roadmap layer — PAR (parallel adversarial review) audits, an iteration log, a behavior-scenario corpus. The correction notes that layer existed but was illegible to a reader who never opened it — which preserves your surface-legibility point while fixing the "no reviewer" claim.
Thanks for the rehab — it made the repo something a stranger can actually build.