sourcecode 5.8.23__tar.gz → 5.8.24__tar.gz

{sourcecode-5.8.23 → sourcecode-5.8.24}/CHANGELOG.md

@@ -1,5 +1,65 @@
 # Changelog
 
+## [5.8.24] — 2026-08-22
+
+*The thirteenth audit-pass queue is closed entire: all six rows as separate
+commits, the regression first, with the full suite green at 12 329 tests. Three
+of the reported mechanisms are wider than reported and are corrected against
+measurement; one defect was found by the battery rather than by either.*
+
+### Fixed — the regression the pass led with
+
+- **`cache warm` warms every root view (`AUD-600-R01`)** — the warm built the
+  compact view only, so the first `--agent` call after an explicit warm paid a
+  full analysis: measured on `mall` (524 Java files) `--agent` 7.59 s → 0.34 s
+  (×22) and the raw contract view 3.20 s → 0.33 s (×9.7), against `--compact`
+  flat at 0.34 s. The field measured ×49.2 at ~3 000 files. Each view turns a
+  different set of analysers on and therefore has its own L1 core;
+  `cache_model.ROOT_VIEWS` is now the one authority the warm iterates and
+  `cache model` publishes a row per view. `--views` warms a subset, and a view
+  left cold is named rather than discovered by its first call.
+
+### Fixed — the new pack surface
+
+- **The gate composes `verify-edit` (`AUD-600-B01`)** — a pack named *Change
+  Safety Gate* did not compose the change-safety command, so a working tree
+  that had just gained `GET /internal/dump-config` returning `System.getenv(key)`
+  answered `NO_CHANGES` / exit 0. It now answers `UNVERIFIED` / exit 2 with the
+  route and its `access_policy`. The general half is the rule: every pack
+  publishes a `non_coverage` block, in its payload and in `pack list`, and the
+  battery asserts it over the axis and rule catalogues.
+- **`--compact` bounds the assessment pack (`AUD-600-B02`)** — the flag was
+  forwarded to the children and nothing else. It now substitutes each
+  component's own decision summary for its body: `mall` 608 K → 42 K estimated
+  tokens, `thingsboard` 8.5 K, both under the ceiling. A refusal may not deny a
+  variant the command declares, so `--format markdown` is derived as the
+  bounded rendering it is.
+- **The report keeps the figures the pack carries (`AUD-600-B03`)** — three
+  cells read `unknown` while the figure sat in a component payload the same
+  pack had written. One catalogue, `MARKDOWN_FIGURES`, read with the same
+  reader the bounded projection uses; every remaining `unknown` names what it
+  looked for.
+- **A write refusal is decided from `argv` (`AUD-600-D01`)** — `--output-dir`
+  inside a read-only tree cost 47.9 s before refusing. The destination is now
+  decided before any component runs and the run is served without its artefacts,
+  the shape `verify --init --no-write` established. Nothing is created inside
+  the analysed tree and every refusal is still announced with its path and type.
+
+### Added
+
+- **`ask schema packs-v1` (`AUD-600-D02`)** — what each pack composes, what it
+  does not, its verdict vocabulary, its exit codes and the aggregation rule that
+  produced the verdict, generated from the pack declarations. The rule was
+  inferable from one run and published nowhere.
+- **Per-component timing in the pack manifest** — `duration_ms` per component
+  and a `timing` rollup naming the slowest one, so a 47 s gate can be attributed
+  from outside. A clock is never an answer: `timing` is excluded from
+  `content_id`, or two runs of an unchanged tree would disagree.
+- **The fourth exit code for packs** — when every unverified component was cut
+  short by its budget the pack leaves with `75` and names them in
+  `components_cut_short`. A mixed run keeps the stronger `2` and the verdict
+  word never moves.
+
 ## [5.8.23] — 2026-08-22
 
 *The external corpus follow-up is closed: pack output is durable by default,

{sourcecode-5.8.23 → sourcecode-5.8.24}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sourcecode
-Version: 5.8.23
+Version: 5.8.24
 Summary: Persistent structural context and ultra-fast repeated analysis for AI coding agents
 Project-URL: Homepage, https://github.com/HarounDominique/sourcecode
 Project-URL: Repository, https://github.com/HarounDominique/sourcecode
@@ -58,7 +58,7 @@ Description-Content-Type: text/markdown
 > question. All analysis is local and deterministic; see [Privacy](#configuration--privacy)
 > before inspecting a repository whose telemetry policy you need to control.
 
-![Version](https://img.shields.io/badge/version-5.8.23-blue)
+![Version](https://img.shields.io/badge/version-5.8.24-blue)
 ![Python](https://img.shields.io/badge/python-3.9%2B-green)
 
 Copyright (c) 2026 Dominique Haroun. All rights reserved. ASK Engine 5.7.0 and
@@ -176,7 +176,7 @@ set `ASK_MAX_RSS_MB=<megabytes>`; if the process exceeds it, ASK refuses the
 answer with `MEMORY_TOO_LARGE`, including the observed peak and limit. Leave it
 unset, or set it to `0`, to disable the ceiling explicitly.
 
-**Current performance status.** Release 5.8.23 retains the shared Canonical IR
+**Current performance status.** Release 5.8.24 retains the shared Canonical IR
 path used by `validation` and now passes that same CIR into labelled `data-exposure`
 runs; the no-label path remains parse-free. The one-model-per-declared-label path is
 also retained. These are implementation properties, not duration promises:
@@ -210,7 +210,7 @@ brew tap haroundominique/sourcecode && brew install sourcecode
 # pip / pipx
 pipx install sourcecode        # or: pip install sourcecode
 
-ask version                    # ask 5.8.23 — and, on a build that has aged,
+ask version                    # ask 5.8.24 — and, on a build that has aged,
                                # how many releases have probably shipped since
 ```
 

{sourcecode-5.8.23 → sourcecode-5.8.24}/README.md

@@ -10,7 +10,7 @@
 > question. All analysis is local and deterministic; see [Privacy](#configuration--privacy)
 > before inspecting a repository whose telemetry policy you need to control.
 
-![Version](https://img.shields.io/badge/version-5.8.23-blue)
+![Version](https://img.shields.io/badge/version-5.8.24-blue)
 ![Python](https://img.shields.io/badge/python-3.9%2B-green)
 
 Copyright (c) 2026 Dominique Haroun. All rights reserved. ASK Engine 5.7.0 and
@@ -128,7 +128,7 @@ set `ASK_MAX_RSS_MB=<megabytes>`; if the process exceeds it, ASK refuses the
 answer with `MEMORY_TOO_LARGE`, including the observed peak and limit. Leave it
 unset, or set it to `0`, to disable the ceiling explicitly.
 
-**Current performance status.** Release 5.8.23 retains the shared Canonical IR
+**Current performance status.** Release 5.8.24 retains the shared Canonical IR
 path used by `validation` and now passes that same CIR into labelled `data-exposure`
 runs; the no-label path remains parse-free. The one-model-per-declared-label path is
 also retained. These are implementation properties, not duration promises:
@@ -162,7 +162,7 @@ brew tap haroundominique/sourcecode && brew install sourcecode
 # pip / pipx
 pipx install sourcecode        # or: pip install sourcecode
 
-ask version                    # ask 5.8.23 — and, on a build that has aged,
+ask version                    # ask 5.8.24 — and, on a build that has aged,
                                # how many releases have probably shipped since
 ```
 

{sourcecode-5.8.23 → sourcecode-5.8.24}/docs/CACHE.md

@@ -18,12 +18,14 @@ Before optimising a workflow, ask the model that ships with the installed CLI:
 ```bash
 ask cache status .       # what this repository already has
 ask cache model .        # whether a warm helps each command
-ask cache warm .         # compact view plus shared structural layers
-ask cache warm . --agent # add the agent view when that is what the workflow reads
+ask cache warm .                 # every root view plus shared structural layers
+ask cache warm . --views compact # one view only, when that is all the workflow reads
 ```
 
-`cache warm` is not a general acceleration switch. It warms the compact view and shared
-layers for the exact tree and scope you name. A command may receive the finished answer,
+`cache warm` is not a general acceleration switch. It warms every declared view of the
+root analysis and the shared layers for the exact tree and scope you name — each view
+has its own core, so a warm that built one left the others paying a full analysis on
+their first call (AUD-600-R01); the warm names the views it built and any it skipped. A command may receive the finished answer,
 reuse only shared work, or receive no benefit; `cache model` names which. Any analysed-file
 change, committed or not, creates a different cache identity. In CI, persist the ASK cache
 directory only when the job policy allows it; otherwise budget a cold run and use
@@ -79,13 +81,21 @@ the enclosing monorepo, and `cache status` / `cache clear` resolve to that same
 <!-- BEGIN GENERATED: ask cache model --markdown -->
 | Layer | What it stores | What invalidates it | `cache warm` | Written by |
 |---|---|---|---|---|
-| `snapshot` | the rendered answer of the root command (L1 core analysis + L2 view) | tree state · analyzer fingerprint · analysis flags (--env-map, --depth, --docs, --exclude, …) for the core; presentation flags (--compact, --agent, --full, --format, …) for the view | built for the compact view (`--agent` also builds the agent view) | the root analysis alone — `ask <repo>` and `ask cache warm`. AUD-591-Q01: `cache_layers.snapshot` read `cold` in all 240 field invocations across 16 repositories, including after `cache clear --all -y` plus two runs, because every one of those runs was of another command. No other command writes `core-*.json.gz`, so on any run but the root analysis this layer is `cold` by construction and its being cold says nothing about the cache. |
+| `snapshot` | the rendered answer of the root command (L1 core analysis + L2 view) | tree state · analyzer fingerprint · analysis flags (--env-map, --depth, --docs, --exclude, …) for the core; presentation flags (--compact, --agent, --full, --format, …) for the view | built for every declared root view (compact, agent, standard) — one core each, because each view turns a different set of analysers on | the root analysis alone — `ask <repo>` and `ask cache warm`. AUD-591-Q01: `cache_layers.snapshot` read `cold` in all 240 field invocations across 16 repositories, including after `cache clear --all -y` plus two runs, because every one of those runs was of another command. No other command writes `core-*.json.gz`, so on any run but the root analysis this layer is `cold` by construction and its being cold says nothing about the cache. |
 | `ris` | the Repository Intelligence Snapshot — structural index, endpoint index, summaries | tree state (the snapshot records the tree it describes) | rebuilt on every warm | every command that reads it |
 | `cir` | the shared Canonical IR — the Java parse every knowledge command reuses | tree state · analyzer fingerprint · schema version | built | every command that reads it |
 | `task` | one `prepare-context` task answer, per task and per option set | tree state · task · --symptom / --all / --include-config / --format | never — a warm does not run any task | every command that reads it |
 | `timeline-samples` | one watched metric as measured at one sampled commit, so a series re-run over an overlapping range measures only what is new (C3-129) | the analyzer fingerprint and the commit sha, both in the key. The tree at a sha never changes, so an entry is never stale — the fingerprint is what keeps a corrected analysis from being served from a previous release. `timeline --no-cache` (or `ASK_TIMELINE_NO_SAMPLE_CACHE=1`) measures every sample again | never — a warm analyses the working tree, not history | every command that reads it |
 | `parse` | the symbol extraction of one file, keyed by its bytes | the file's own bytes and the extractor's source — content-addressed, so it is never stale and never needs invalidating. Bounded by `ASK_PARSE_CACHE_MAX_MB` (default 512 MB, least-recently-used, swept every 32 MB written rather than on every write, and the last interval before a process exits is never swept — so the store rests up to one interval per writing process above its budget, and the figure to size a disk by is the ceiling `cache status` publishes beside it, not the budget itself (ASK-15)) and reported in `cache status` (F-AR) | filled for every Java file the warm parses | every command that reads it |
 
+| Root view | Invocation | Its own core | `cache warm` |
+|---|---|---|---|
+| `compact` | `ask <repo> --compact` | turns dependencies, env map, code notes and architecture on; bounded, and the view the front page recommends first | builds it |
+| `agent` | `ask <repo> --agent` | everything the compact core has plus the module import graph, and it drops the file tree — a different core, not a different rendering. This is the view the root help recommends to AI agents | builds it |
+| `standard` | `ask <repo>  (and `ask <repo> --full`, which is this view with the display caps raised)` | the raw contract view: none of the compact analysers run, so its core is neither of the two above and a warm of either leaves it cold | builds it |
+
+Each view turns a different set of analysers on, so each has its own L1 core: a warm that builds one leaves the others paying a full analysis on their first call (AUD-600-R01).
+
 Measured on BroadleafCommerce (2 000+ Java files, dirty tree), warm machine, 3.2.2, each command in isolation.
 
 Measured on release 3.2.2. Figures are that build's on that repository, not a prediction for yours; the nightly performance gate (`docs/perf/REGRESSION-GATE.md`) is what re-measures them.
@@ -96,7 +106,7 @@ The second anchor is the field one, and it carries its own build for the same re
 |---|---|---|---|---|---|
 | `ask (root)` | the answer | --compact 13.3 s cold → 0.3 s warm (cold re-measured on 3.7.0: was 19.3 s, C3-6); --agent --full --env-map --depth 20 34.7 s → 33.9 s (no gain) | yes | `snapshot`, `ris`, `parse` | `--compact` is what a warm stores by default; `--agent` needs `cache warm --agent`. `--env-map`, `--depth N` and `--exclude` change the *analysis*, so they miss the warmed core and rescan — the 171 s the field measured after a 103 s warm on 4.10.4. Eval #23 timed the same invocation at 72,3 s from a purged cache, and then paid 71,7 s again for `--agent` on the state it had just analysed (C3-103). |
 | `posture` | the shared work | 10.1 s → 1.6 s | no | `cir`, `parse` | Resolves the conditional bean graph on every run, over the shared CIR a warm builds — the parse it used to repeat for itself. `--diff` compares two profile sets over that one IR, so the second side costs the resolution only. |
-| `risk` | the shared work | not measured on the battery yet — the composition is bounded by the `spring-audit` + `impact-chain` costs listed here, not by new analysis | no | `cir`, `parse` | Composes what the audit, impact-chain and the posture already answer, so it pays each of their costs once over the shared CIR a warm builds — one parse for the whole composition, and the reachability query is cached per symbol within the run. |
+| `risk` | the shared work | 240.1 s on the audited monolith (13th pass, rc 0, 178 171 B). The two rounds before it exhausted every budget they gave it — 301 s under a 300 s budget (5.8.20) and 401 s under 400 s (5.8.22) — and the refusal with its budget is the honest anchor for a run that does not finish, never a smaller number measured somewhere else | no | `cir`, `parse` | Composes what the audit, impact-chain and the posture already answer, so it pays each of their costs once over the shared CIR a warm builds — one parse for the whole composition, and the reachability query is cached per symbol within the run. |
 | `enrich` | the shared work | not measured on the battery yet — bounded by the `risk` composition, plus reading one JSON file | no | `cir`, `parse` | Runs the same composition as `risk` over the repository, then joins a SARIF log to it. Reading the log is negligible; everything a warm helps with is the repository side, so what it buys is what it buys `risk`. |
 | `audit-report` | the shared work | not measured on the battery yet — bounded by `risk` + `posture`, or by `posture` when `--from-risk` is supplied | no | `cir`, `parse` | Packages `risk` and `posture` evidence. A warm helps the repository side; `--from-risk risk.json` skips risk recomputation entirely and only builds the posture/report projection. |
 | `migrate-recipe` | the shared work | not measured on the battery yet — bounded by `migrate-check` on the same repository (openmrs-core ~2 s) | no | `parse` | Runs the same scan as `migrate-check` and projects its findings into an OpenRewrite recipe, so it buys exactly what a warm buys `migrate-check`: the parse, not the rule pass. |