Repository navigation
Conversation
Add reproducible Terraform plan, state, and graph fixtures covering startup and shutdown rules, explicit and inferred phases, nested and empty modules, `count`, `for_each`, overlapping rules, and unordered scripts. Verify that plan conversion validates rules without attaching runtime dependencies, while post-apply conversion produces the expected ordered graphs and per-script dependencies. Check diagnostics in both paths. Add selector coverage for all matching managed `for_each` instances. Refs: https://linear.app/codercom/issue/PLAT-552
|
/coder-agents-review Please limit the review to P0, P1, and P2 issues only. |
|
Chat: Review posted | View chat Review history
deep-review v0.13.0 | Round 1 | Last posted: Round 1, no findings, COMMENT. Review Finding inventoryFinding inventory: PR #31038Findings
Law analysisRound 1. Effective LOC +2234 (1852 Terraform-generated, 382 hand-written). Head 7b5b606. Verdict: Don't split. Enforcement: Advisory. Contested and acknowledgedRound logRound 1Full panel (round 1). Focused (Go tests and Terraform fixtures only): ging-go, ryosuke, pariston, mafuuu, bisky, komugi, gon, leorio, wildcard knov. Netero: 1 P3, 1 Note. Law: don't split. Cross-check: Gon rated CRF-4 and CRF-5 P2. Keep-argument: AGENTS.md requires comments to describe behavior, and CRF-4's comment partly narrates the test. Downgraded to Nit because both comments are accurate and the change is wording only, with no behavior or reader-misleading consequence. CRF-3 converges Pariston (P3) and Knov (Note); higher wins at P3. Keep-at-P3 argument: About deep-reviewCRF = Coder Review Finding (P0-P4, Nit, Note)
|
There was a problem hiding this comment.
This PR adds Terraform-generated script-order fixtures, a plan and state conversion test that uses them, and a for_each selector case. The review found no P0, P1, or P2 issues. As requested, lower-severity findings are not posted. Both new tests pass at 7b5b606.
Out of scope (needs a ticket or explicit acceptance by a human):
provisioner/terraform/resources.go:1166: on the plan path,dependsOnAgentmatches only a directcoder_script.<name>tocoder_agentgraph edge, so module-declared scripts are attached to no agent. Plan-source conversion of this PR's fixture keeps 4 of its 8 scripts, and the existingtestdata/resources/module-scriptplan golden also has no module script.
🤖 This review was automatically generated with Coder Agents.
Add reproducible Terraform plan, state, and graph fixtures covering
startup and shutdown rules, explicit and inferred phases, nested and
empty modules,
count,for_each, overlapping rules, and unorderedscripts.
Verify that plan conversion validates rules without attaching runtime
dependencies, while post-apply conversion produces the expected
ordered graphs and per-script dependencies. Check diagnostics in both
paths.
Add selector coverage for all matching managed
for_eachinstances.Refs:
https://linear.app/codercom/issue/PLAT-552