Skip to content

Claude/evaluate monorepo structure l lwpe#100

Merged
Exelord merged 3 commits into
mainfrom
claude/evaluate-monorepo-structure-lLWPE
May 17, 2026
Merged

Claude/evaluate monorepo structure l lwpe#100
Exelord merged 3 commits into
mainfrom
claude/evaluate-monorepo-structure-lLWPE

Conversation

@Exelord
Copy link
Copy Markdown
Member

@Exelord Exelord commented May 17, 2026

No description provided.

claude added 3 commits May 16, 2026 13:21
We re-monorepo'd in 2026-05 (PR #25) only to host the dashboard UI's
Vite build alongside the runner; once the dashboard was deleted the
monorepo lost its reason to exist and we flattened back. This doc
audits the current shape, looks at four restructuring options, and
recommends staying flat — the seven src/ subdirs already form a clean
DAG and per-file module docs cover discoverability.

Concrete follow-ups proposed (not done here): codify the boundaries
as an oxlint no-restricted-imports rule (the code already respects
every edge), delete the untracked apps/ + packages/ debris dirs, and
note the decision in CLAUDE.md so the next agent doesn't re-litigate.
`patterns.md` is the inverse of `comparison.md` (which lists gaps):
subsystem-by-subsystem tables of what we deliberately inherited
from Turbo or Nx, every row anchored to a source-file citation.
`grep -rn "Turbo\|Nx parity\|turborepo" src/` returns ~20 explicit
parity comments — the doc indexes each one so a reader can audit
"is this really how Turbo does it" without a tour.

`benchmarks.md` captures the latest 100-project × 3-task numbers:
vx adds 2.95 s of overhead vs raw shell on cold runs and 159 ms
on full restore, against Turbo's 11.38 s / 589 ms and Nx's
20.62 s / 858 ms. Includes a per-task overhead breakdown, a
"why it's fast" section pointing at the architectural choices
that move the needle (no daemon, xxHash3, hardlink restore,
in-process tar parser, Bun I/O), and a headroom section listing
the next things worth profiling.

README index gains two rows for the new pages.
PR #100's CI failed on format-check because the new docs (patterns.md,
benchmarks.md) used inconsistent column padding in markdown tables.
PR #96 / #97's monorepo-structure.md regressed the same way.
`bun src/bin.ts run format` rewrites all three to oxfmt's canonical
table alignment. No content changes.
@Exelord Exelord merged commit eb01ec4 into main May 17, 2026
1 check passed
Exelord pushed a commit that referenced this pull request May 17, 2026
PR #100's CI failed on format-check because the new docs (patterns.md,
benchmarks.md) used inconsistent column padding in markdown tables.
PR #96 / #97's monorepo-structure.md regressed the same way.
`bun src/bin.ts run format` rewrites all three to oxfmt's canonical
table alignment. No content changes.
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.

2 participants