Skip to content

docs(pages): point every source link at nnunley/fmpl [DRAFT] - #2

Draft
mparrett wants to merge 1 commit into
mainfrom
fix-fork-urls
Draft

mparrett wants to merge 1 commit into
mainfrom
fix-fork-urls

Conversation

@mparrett

@mparrett mparrett commented Jul 25, 2026 •

Copy link
Copy Markdown
Collaborator

Source-link sweep, deferred pending #4 (where the canonical home for these pages is). The change itself is mechanical and ready; which direction the links point depends on that answer.


The published pages carry fork URLs in their mastheads, footers, and the clone command — a holdover from being written on my fork before you merged the work. Right now nnunley.github.io/fmpl sends readers to github.com/mparrett/fmpl for the source, which is the wrong place to send them (and that fork is in flux).

Audited nine references across four pages, all mechanical:

  • docs/index.html — masthead eyebrow, footer source link
  • docs/fmpl-tour.html — masthead eyebrow, git clone command, footer
  • docs/fmpl-field-log.html — masthead eyebrow, footer
  • docs/fmpl-rehab-log.html — masthead eyebrow, footer

The (fork of nnunley/fmpl) suffix could go away with them, since it no longer says anything.

The acknowledgement in README.md is untouched — that's a link to a person, not to a source repo. (also: Thank you 🙇🏻 )

I split out from the language-guide PR so it can land on its own; the two don't conflict in either order.

Heads up- this touches docs/fmpl-rehab-log.html, which your open #1 also edits. No overlap in this PR — yours changes prose in §2 and §7, this only touches the masthead and footer — and I verified the two merge clean in either order. Happy to rebase on #1 if you'd rather land that first.

Companion to #3, which adds the language guide.

The published pages carried fork URLs in their mastheads, footers, and
clone command — a holdover from being written on the fork before this
work was merged upstream. They now point readers at the canonical repo.

The acknowledgement in README.md is a profile link, not a source link,
and is left alone.
@mparrett
mparrett marked this pull request as draft July 25, 2026 22:23
@mparrett mparrett changed the title docs(pages): point every source link at nnunley/fmpl docs(pages): point every source link at nnunley/fmpl [DRAFT] Jul 25, 2026
@mparrett
mparrett requested a review from nnunley August 6, 2026 13:42
chazu pushed a commit to chazu/fmpl that referenced this pull request Oct 5, 2026
…blocks

Discovered while building the doctest harness (fork issue nnunley#2), by running
every fenced block through eval() instead of the line-at-a-time REPL:

- TUTORIAL: a list literal opening a line after a complete expression parses
  as *indexing* of that expression (the parser is newline-insensitive), so
  stacked literals were parse errors as written. Terminate with `;` — a
  trailing semicolon preserves the expression's value. Documented as Issue 5
  in specs/parser-limitations.md.
- DEMO: the multiple-cursors example called `cursor1::position(...)` /
  `cursor2::position(...)`, but `cursor::` is a builtin namespace, not a
  method-on-variable syntax — `cursor1::` resolves to null and errors. Use
  `cursor::position(cursor1)` etc.
- TUTORIAL: network-dependent examples (curl) are marked
  `<!-- fmpl-doctest: skip -->`; design sketches (recursive agent loop,
  grammar-based agents, checkpoint/resume) now use the ```fmpl-sketch tag.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
chazu pushed a commit to chazu/fmpl that referenced this pull request Oct 5, 2026
Closes nnunley#2. Locks in the 2026-07-20 REPL sweep: fenced ```fmpl blocks are
extracted and evaluated, and `-- Returns:` / `-- =>` / `// =>` annotations
are asserted against the result's display form, so a documented result that
drifts from real behavior fails CI.

Per annotated line, the block prefix through that line runs in a fresh VM;
every block is also evaluated in full and must not error. Trailing
parentheticals in annotations are explanatory prose; %{...} displays compare
order-insensitively at top level (Value::Map is a HashMap). Opt-outs:
<!-- fmpl-doctest: skip --> before a fence, or the ```fmpl-sketch tag.
Conventions noted in AGENTS.md, full detail in the harness header.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
chazu pushed a commit to chazu/fmpl that referenced this pull request Oct 5, 2026
… tour counts

New docs/fmpl-field-log.html, a companion page to the tour in the same
visual language: a retrospective on the nnunley#2/nnunley#3/nnunley#4 arc — the five stacked
bugs behind the token-738 error, the fork points (postlude vs orphan
module, PEG backtracking vs the legacy 20-char heuristic, EOF enforcement,
the @ operand restriction, env-var routing), and the 98-test burn-down
that collapsed to two missing definitions.

Tour updates for staleness the same work caused: test counts 1519/74,
epoch 9, 7 crates (fmpl-wasm card added), gap-ledger metacircular row
marked closed with a link to the field log, footer dated.

Verified with Playwright in light and dark themes: no horizontal
overflow, no console errors, links resolve.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant