Repository navigation
#268 — vars change tiers: precedence-aware attribution + honest blast-radius copy - #287
Conversation
…l entries, reference chips, no widening
Per changed dbt_project.yml var, the project-change panel row now
carries precedence-resolved entries (fusion's order mirrored exactly:
CLI --vars > package-scoped vars > global vars > inline var() default,
the default LOWEST) with affected-model lists at honest evidence tiers:
- DIRECT — var('x') in a Jinja-evaluated region of the model's
raw_code (call grammar exhaustive per fusion var.rs:25-92,111-126,
incl. var.has_var and map defaults; {# #} comments and {% raw %}
blocks never count)
- CONFIG — recursive unrendered_config scan (braced and bare
expression strings; covers project-file, schema-YAML, inline
config() and hooks)
- MACRO — depends_on.macros transitive closure over manifest macro
bodies, the mediating macro named per hit
- UNKNOWN residual — in-row copy enumerating the causes (dynamic
names, var-to-var indirection, CLI --vars masking, Python models),
stating what WAS checked, 'at least N' framing, the
disabled-membership caveat, and the contextualize-don't-widen
statement
A global edit is masked per package pinned by an unchanged
package-scoped value; unit tests pinning the var in overrides.vars are
reported insulated (the override always wins). R1b cap: past 10 names
a tier list collapses to its count with the full list in a <details>
('listed, not individually rendered'). In-scope models referencing an
edited var get var-reference chips beside the #267 config chips —
context only, the in-scope set NEVER widens on a vars edit (founder
v3: contextualize-don't-widen).
Build gate (the #145 lesson): non-empty macros[].depends_on verified
on REAL fixtures from BOTH engines — dbt-core 1.11 (jaffle 324/485,
playground 639/910 committed fixtures) and dbt-fusion
2.0.0-preview.177 (334/510, local dogfood compile); dispatch
indirection confirmed resolved in both.
Dogfood: the diff-showcase patch gains the second vars hunk
(default_state_grid_density weekly→monthly) and the synthetic
playground manifest gains a DIRECT reader (mart_dq_summary), a
CONFIG-driven node (mart_date_state_grid), a MACRO-mediated pair (via
add_dq_flags), and the override-pinned unit test matching the
committed source YAML — all four tiers plus the insulation note are
visible in examples/diff-showcase-report.html, and both in-scope
models carry their reference chips.
The 8-theme AA matrix guard extends to the three new tier-chip classes
(56 measurements); all goldens + the chrome snapshot regenerated over
the merged tree, intended-only audited.
Closes-by-PR: cute-dbt#268 (epic #262, slice C3).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…on, type alias, trailing comma) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
Warning Review limit reached
More reviews will be available in 25 minutes and 50 seconds. Learn how PR review limits work. Your organization has run out of usage credits. Purchase more credits in the billing tab to continue. ⌛ How to resolve this issue?After more reviews become available, a review can be triggered using the We recommend that you space out your commits to avoid hitting the rate limit. 🚦 How do rate limits work?CodeRabbit enforces hourly rate limits for each developer per organization. Our paid plans include higher PR review limits than trial, open-source, and free plans. In all cases, reviews become available again over time. During sustained high-volume PR review activity, CodeRabbit may temporarily slow when the next review becomes available. Please see our Fair Usage Limits Policy for further information. ℹ️ Review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: ⛔ Files ignored due to path filters (1)
📒 Files selected for processing (20)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
📄 Rendered report previewAll golden examples regenerated cleanly. 🟡 Golden examplesCommitted to
🐶 Live dogfood previewThis PR doesn't touch ▶ Open ↗ opens the report in your browser in one click — The Pages preview may take ~1 min to update after this comment Alternative: GitHub CLI# gh CLI >= 2.63 extracts into ./report-preview-playground/.
gh run download 27404482074 -R breezy-bays-labs/cute-dbt -n report-preview-playground
open report-preview-playground/playground-report.htmlPosted by |
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request implements static vars change attribution at honest tiers (DIRECT, CONFIG, MACRO, and UNKNOWN residual) in cute-dbt. It introduces a static scan over the manifest to resolve precedence-aware var edits and displays them in the project definition panel with tiered affected-model lists, package masking, and unit test insulation. It also adds var-reference chips to in-scope models in the report. The reviewer identified a compilation issue in src/domain/vars.rs where OsStr is used with eq_ignore_ascii_case, which is not implemented for OsStr, and provided a suggestion to convert it to a &str slice first.
Important
The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.
Overnight orchestration wrap-upMerged as What shipped (#262 C3, the big-batch slice): the vars tier scanners (DIRECT via fusion's exhaustive var() call grammar with Jinja-region guarding; CONFIG via recursive unrendered_config walk; MACRO via depends_on transitive closure with the mediating macro named), fusion's var precedence pinned by total-order property tests (CLI > package > global > inline default LOWEST), the locked honest blast-radius copy rendered verbatim in the golden (in-row, enumerated causes, "at least N", what WAS checked), the R1b cap shared with #267, override-insulated unit tests subtracted, and zero scope widening — vars contextualize, never widen, per the founder-locked decision. This completes the #262 lane's buildable arc for the night: C1 (#266) + C2 (#267) + C3 (#268) + C4 (#269) all merged. Remaining: C5 (#270, explorer pane — now unblocked) and the macro epic (#265, needs shaping). 🤖 Generated with Claude Code |
1 participant
Summary
Closes #268 (epic #262, slice C3 — the big-batch differentiator: fusion's own
state:modifiedcannot see var-value edits).Per changed
dbt_project.ymlvar, the project-change panel's vars row now carries precedence-resolved per-var entries with affected-model lists at honest evidence tiers, plus reference chips on in-scope models. A vars edit never widens the in-scope set (founder v3: contextualize-don't-widen).Precedence (fusion's, mirrored exactly)
CLI --vars > package-scoped project vars > global project vars > inline var() default(ConfiguredVar::call_as_function,configured_var.rs:55-129@9977b6cb…). The inline default is LOWEST — a call-site default never overrides a project value, so carrying one never subtracts a model. CLI--varsis unobservable → an explicit in-row caveat, never a guess. Root-project vars extend into every package's namespace (load_vars.rs:28-42); a package-scoped pin masks a global edit for exactly that package (stated per row). Package-scope keys invars:expand per inner var; the total order is pinned by exhaustive property tests.Tiers (per AC#1, the research evidence)
var('x')in a Jinja-evaluated region of the model'sraw_code. Call grammar exhaustive per fusionvar.rs:25-92,111-126:var('x')/var("x"), positional +default=kwarg defaults incl.Noneand map defaults (brace/string-aware region ends so a dict default never truncates), andvar.has_var('x').{# #}comments,{% raw %}blocks, plain SQL text, and lookalikes (myvar(,varchar(,graph.var() never count; a SQL--comment inside{{ }}does (Jinja evaluates it — true positive).unrendered_configscan: braced strings (project-file / schema-YAML sources) and bare expression strings (fusion's inline-config()raw-text preservation,utils.rs:713-721).depends_on.macrostransitive closure over manifest macro bodies; each hit names its mediating macro (stg_enc (via add_dq_flags)).var()names, var-to-var value indirection, CLI--varsoverrides, Python models), what WAS checked stated (Checked: 34 models' SQL and configs plus 89 macro bodieson the showcase), "at least N" framing, the disabled-membership caveat, and the never-widened statement. Dynamicvar(expr)callers surface as a cannot-rule-out bucket;.pymodels skip the SQL scan and are counted honestly.Override-pinned unit tests are subtracted: fusion re-binds
varwithoverrides.varsat unit-test compile (unit_test.rs:603-609) — pinning tests are listed as insulated.R1b cap (AC#3)
Past 10 names a tier list collapses to its explicit count — "at least 47 models read this var directly in SQL — listed, not individually rendered" — with the full list in a collapsed
<details>. Same cap constant as the #267 config-tree rows.Build gate (AC#4, the #145 lesson)
Non-empty
macros[].depends_onverified on REAL fixtures from BOTH engines before relying on the MACRO tier:macro.dbt.current_timestamp → macro.dbt_duckdb.duckdb__current_timestamp).Dogfood (visible in the golden showcase)
examples/diff-showcase-report.htmlnow renders, byte-gated:dq_quarantine_threshold(10 → 5): DIRECTmart_dq_summary, MACROstg_synthea__encounters+stg_synthea__medications(viaadd_dq_flags), and the insulatedtest_mart_dq_summary_zero_quarantined_when_all_valid(overrides now in the synthetic manifest, matching the committed source YAML);default_state_grid_density(weekly → monthly, the new second vars hunk): CONFIGmart_date_state_grid;Intended side effect of the fixture edit: in baseline-mode
examples/playground-report.htmlthe dq unit test now readschanged: true(its overrides differ from the baseline manifest — a truthful struct diff).Tests
src/domain/vars.rsincl. house-style exhaustive property tests: precedence total order over every presence combination, package-vs-global resolution, masking arms (pin-both-sides / pin-only-new), scanner call-form grammar,changed_varsunion semantics + scope expansion edge cases.tier-chipclasses (56 measurements; recipes reuse the existing AA-gated token pairs).Gates (all run directly, not via lefthook-in-fresh-worktree)
cargo fmt --check·clippy --all-targets --locked -D warnings(incl. the new-stable lints the pre-push hook surfaced) ·nextest --all-targets --locked1527 passed · BDD 23 features / 182 scenarios passed · headless pair (headless_toggle -- --ignored83 passed,headless_zero_egress -- --ignored11 passed) ·cargo docboth modes-D warnings·cargo deny check· crap4rs scorecard PASS (0 above threshold, worst 24.1) · goldens + chrome snapshot regenerated over the merged tree (post-#286) with intended-only audit · fixture MANIFEST SHAs updated.🤖 Generated with Claude Code