Skip to content

chore(dependabot): weekly npm, docker and actions updates - #1

Open
devdudeio wants to merge 1 commit into
mainfrom
chore/dependabot-config
Open

devdudeio wants to merge 1 commit into
mainfrom
chore/dependabot-config

Conversation

@devdudeio

Copy link
Copy Markdown
Member

Adds .github/dependabot.yml. The repo had none.

Ecosystems (detected from the tree, with the monorepo layout in mind):

  • npm at / - pnpm workspace (pnpm-workspace.yaml -> packages/*,
    apps/*) with one pnpm-lock.yaml. A single root entry covers every
    workspace package; per-package entries would contend for the same lockfile.
  • docker at /docker - demo-server.Dockerfile,
    facilitator.Dockerfile, proxy.Dockerfile, all FROM node:26-slim.
  • github-actions at / - ci.yml, release.yml (SHA-pinned).

Style follows chainvue/verus-rpc (same semantic-release angular preset):
weekly, fix/chore/ci prefixes with include: scope. The npm limit
is 10 rather than 5 because this is a 16-package workspace.

Why the groups are cut this way. The trigger: in peculium-wallet Dependabot
bumped vitest and left @vitest/coverage-v8 on the old major, and the peer
mismatch broke CI. Grouping is the only way to force a peer set into one PR, so
every peer family in this repo gets one:

  • vitest - vitest, @vitest/*, vite (all update types, majors
    included - the exact case that broke).
  • eslint - eslint, @eslint/*, typescript-eslint,
    @typescript-eslint/*, globals, typescript.
  • nestjs - @nestjs/* + reflect-metadata + rxjs. These peer-depend
    on each other and are spread across facilitator, proxy, mcp, adapter-nestjs
    and demo-server; a half-applied major would break all of them at once.
  • drizzle-sqlite - drizzle-orm, drizzle-kit, better-sqlite3,
    @types/better-sqlite3 (the kit generates against the orm's schema API, and
    the types package has to match its runtime).
  • semantic-release - core, @semantic-release/*, multi-semantic-release.
  • dev-dependencies - everything else dev, minor/patch.

Ignore: typescript majors, same gate as verus-rpc (typescript-eslint peer
range >=4.8.4 <6.1.0 as of 8.6x).

YAML parses and the keys were checked against the Dependabot schema. One thing
I could not verify from here: Dependabot's Docker file matcher and the
<name>.Dockerfile naming. If the first run reports no Docker manifests in
/docker, that entry is the thing to revisit.

https://claude.ai/code/session_01AxFbkWrzkhDieg5RGm8YdK

Detected ecosystems:
- npm at "/" - pnpm workspace (pnpm-workspace.yaml covers packages/* and
  apps/*) with a single pnpm-lock.yaml, so one root entry updates every
  workspace package. Per-package entries would fight over that one lockfile.
- docker at "/docker" - demo-server, facilitator and proxy Dockerfiles, all on
  a node:26-slim tag.
- github-actions at "/" - ci.yml and release.yml, SHA-pinned.

Prefixes follow the repo's semantic-release setup, as in chainvue/verus-rpc.

Groups follow the peer edges, because a split bump is what actually breaks the
build: in peculium-wallet Dependabot bumped vitest and left
@vitest/coverage-v8 on the old major, and the peer mismatch took CI down. This
repo has both that pair and three more of the same shape:
- vitest: vitest, @vitest/*, vite
- eslint: eslint, @eslint/*, typescript-eslint, @typescript-eslint/*, globals,
  typescript
- nestjs: @nestjs/* plus reflect-metadata and rxjs, which they peer-depend on -
  one package moving alone puts the DI container out of step across several
  workspace packages at once
- drizzle-sqlite: drizzle-orm, drizzle-kit, better-sqlite3 and its @types
- semantic-release: core, @semantic-release/* and multi-semantic-release

typescript majors are ignored for the same reason as in verus-rpc:
typescript-eslint's peer range does not admit TS 7, so the PR could not pass
lint.

Claude-Session: https://claude.ai/code/session_01AxFbkWrzkhDieg5RGm8YdK
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