mirror of
https://github.com/colbymchenry/codegraph.git
synced 2026-10-17 05:07:37 +08:00
b50c7c0a70bc666a7855f531b62899500074c8b4
1152 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Colby Mchenry
|
b50c7c0a70 |
fix(swift): a call through a type path lands on the type it names (#2118)
`API.PackageController.GetRoute.query(on:)` kept only `query` — the
receiver was a navigation_expression, not an identifier — and the bare
name exact-matched whichever type's `query` came first. On
SwiftPackageIndex-Server every route type has one, so health checks and
package-controller tests "called" the dependency controller (or a test's
own `query`).
Extraction (TS and the kernel's swift walker, byte-parity): a receiver that
is a path of capitalized type names, two segments or more, not led by
`Self.`, is kept: `API.PackageController.GetRoute.query`. Instance chains
(`self.store.load()`) and mixed paths (`URLSession.shared.data`) are
unchanged.
Resolution (swift-type-visibility.ts, called decisively from
matchReferenceInner so nothing falls through to name guessing): the member
on the type the path names — a method, an enum case with associated values,
or a nested type's initializer. A member's owner path is its qualified name
with the outermost extension's written path read back from its line
(`extension API.PackageController { enum ShowRoute … }` →
`API::PackageController::ShowRoute`), so twin `Feature::State`s declared in
`extension BasicsView.Feature` and `extension ObservableBasicsView.Feature`
are told apart. A path shortened inside its namespace or behind a module
qualifier fits; two owners, or none, leave the call unresolved. TCA's
`@Reducer enum Path { case detail }` generates `Path.State`/`Path.Action`,
so `Path.State.detail(…)` lands on the written case.
A/B vs main (edge diffs, every change classified): SPI 76 retargeted, all to
the type the line names, 7 dropped (Fluent `App.Version.query`, `Plot.Node.*`
— SDK members main pinned on project methods); TCA 61 retargeted, 1 dropped
(`State.StateReducer.scope` → an unrelated `Store::scope`); mastodon-ios 177
retargeted (`Mastodon.API.V2.Instance.instance`, `L10n.Plural.Count.vote`),
11 SDK drops; swift-nio 6 calls into a deprecated alias became instantiates
of `ChannelOptions.Types.*`; IceCubes 1/1, Alamofire 2 SDK drops.
Kernel/wasm parity 0 diffs over 2,752 files on six repos.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
Colby Mchenry
|
e845b1cd10 |
fix(nestjs): routes carry the app's global prefix and URI version (#2117)
`app.setGlobalPrefix('api')` and `app.enableVersioning({ type:
VersioningType.URI, defaultVersion: '1' })` put every controller under
`/api/v1`, but routes were named by the controller path alone (`GET /user`).
A front end's `this.http.get('/api/v1/user')` then linked to nothing:
Ghostfolio's Angular client had no cross-tier edge into its own NestJS API.
postExtract now reads each `NestFactory.create` bootstrap and renames the
routes of controllers in that app, in NestJS's order:
`/{prefix unless excluded}/{v}{version}/{RouterModule prefix}/{path}`.
- `exclude` paths (plain, `{ path, method }`, `{/*wildcard}`, `(.*)`, `*`)
skip the prefix and keep their version, as NestJS does.
- A route's version: its own `@Version` (found past decorators with object
arguments), else the class's `@Version` or `@Controller({ version })`,
else `defaultVersion`. `VERSION_NEUTRAL` has none; header/media-type
versioning adds no segment.
- A computed prefix (`config.get(...)`) is not guessed.
- Each app uses its own bootstrap (the longest root above it). A seed script
that creates the same app without serving it doesn't displace the one that
configures it, and an app with no config doesn't borrow another app's.
- Only routes whose decorator sits in a `@Controller` are touched, so an
Express route in the same repo keeps its name.
Ghostfolio: 0 → 103 cross-tier edges (82 of 85 client call sites), no
mismatches. koel, proshop, IceCubes and nest-samples are byte-identical to main.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
Colby Mchenry
|
c3efbbf4e7 |
feat(angular): a template's event bindings call the component's methods (#2114)
An Angular handler's only caller is its template: `(click)="toggleFavorite()"`,
`(ngSubmit)="submitForm()"`. With no edge, the method showed no callers,
impact stopped at it, and the Steps picture of a screen reached its
handlers only through class containment, with nothing to say what fires
them.
The template pass now reads each `(event)="…"` binding — not
`[(ngModel)]`'s two-way half — and links the component to every method of
its own the statement calls (`save()`, `this.toggle(x)`; not
`closed.emit()` or `form.reset()`), as a `calls` edge
(`synthesizedBy: 'angular-event'`) carrying the binding as
`metadata.trigger`: `{ kind: 'prop', name: '(click)', of: 'button' }`.
The binding is in the template, so it cannot be read back from the source
at the edge's line; Steps now takes a trigger an edge carries before
reading one at the site.
angular-realworld: 21 binding sites → 17 edges (3 repeats of a method and
event, 1 `delete.emit(true)`); Steps for /article/:slug draws
deleteArticle on (click) · <button> → DELETE and back to /, addComment on
(ngSubmit) · <form>, the favorite button's toggle → POST/DELETE or
/register. Ghostfolio 225 edges of 302 sites, ngx-admin 75 of 100 — the
rest are `$event.stopPropagation()`, output emits, assignments and
repeats. Controls (koel, proshop) byte-identical.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
Colby Mchenry
|
761611c18e |
feat(angular): routes, navigation and the component tree for Angular apps (#2113)
An Angular app had no route nodes, no navigates edges and no link from a template to what it renders — the viewer's Steps tab said "No screens or endpoints" and there was no Screens tab. Routes (frameworks/angular-router.ts): every `Routes` array — typed `Routes` / `Route[]`, `RouterModule.forRoot/forChild`, `provideRouter`, a routes file's `export default [...]` / `as|satisfies Routes` — walked as objects. `component` binds by `references` (route-roots takes a class as a named handler); a lazy `loadComponent` binds to the file's `@Component` class. `children` join their parent's path; a lazily loaded file's routes are put under the path that loads them in postExtract (following an NgModule `loadChildren` to the routing module it imports), idempotently from the in-file path kept in the qualified name. A route with children is a layout: a screen only where no child claims its address, and linked from each screen inside it (`references`, `layout: true`). Paths written as route constants (`internalRoutes.account.path`, through destructured locals and `const x + '/:id'`) and `$localize` strings are read from the constant object. `redirectTo` aliases an address to the screen it sends the user to; a root `**` redirect answers `/` only. A `matcher` route and a `**` catch-all are not screens. Navigation: `router.navigate([...])`, `navigateByUrl`, a guard's `createUrlTree` / `parseUrl` — a command array (non-literal elements are holes), a static string, a route constant, a component property holding one, a function constant called with arguments, `.concat(id)`. A `relativeTo` navigation, a query-only `navigate([])` and a destination made only of holes (it would match any route of its length) are left unresolved. Templates (angular-template-synthesizer.ts): a component's `templateUrl` file or inline `template:` is read at synthesis time. `<app-foo>` by element selector (same app first; two candidates = none) is a `calls` edge from parent to child (`angular-template`), the edge a navigation in a child component rides to its screen. `routerLink` / `[routerLink]` and a `routerLink:` field written in the class (a tab bar's config) are navigates edges. Gated on an `@angular/core` dependency. Measured: - angular-realworld: 10 routes all bound; 11/11 navigation sites accounted for (10 resolved, 1 query-only); 31/31 routerLink sites resolved; 18 renders. - Ghostfolio: 84 routes all bound, 0 unresolved constant paths; 27 navigate resolved, 23 correctly not (8 relative, 9 query-only, 2 computed, 4 aimed at a matcher route); 162 routerLink edges, every one checked against its site; 170 renders. - ngx-admin (NgModule): 55 routes, 47 bound (8 are @nebular/auth's); no navigation calls in the source. - Controls byte-identical vs main (nodes and edges incl. metadata): koel, IceCubesApp, spring-petclinic, proshop. Ghostfolio index time at parity. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
7639c787e4 |
fix(swift): what a SwiftUI app's viewer showed that was not so (#2109)
Three things IceCubesApp showed in `codegraph ui`:
Steps labelled calls that leave the index by other stacks' names:
`rawValue.data(using: .utf8)` network (25 boxes), Foundation's `Timer`
telemetry, `Calendar` and a SiriKit `intent` device, the app's own
`Notifications` endpoint enum device (expo-notifications' row), and
`viewModel.votes.firstIndex(of:)` a database read (the `\w*Model` row).
Now: a Swift `.data/.upload/.download/.bytes` is network only on a
`*session` receiver; Timer/Calendar/intent are no effect in Swift; a
`viewModel` is not a table (any language); a keychain receiver is
storage ahead of Swift's generic `x.delete` database row. A call whose
receiver is a type the Swift project declares — not an `extension X {}`,
which declares nothing (read through the site reader) — no longer
matches the library-name rows; the model and reply rows, written for
project types, still do.
"Files that run something" listed every view with a `#Preview { … }`:
the macro sits at a file's top level, so its constructions were
module-level edges. Entry points now subtract calls inside a Swift
file's top-level previews (trailing closures included — WidgetKit's
`} timeline: {`), drop a file left with none, and re-rank.
Every `struct X: View` had a one-line `component` twin, and `@main`
apps, view controllers and UIView subclasses a `class` twin, from the
SwiftUI/UIKit resolvers' regex extract(): 188 nodes on IceCubesApp, 184
on mastodon-ios, zero edges into any of them — dead ends in search and
the symbol view. The two extract()s are removed.
IceCubesApp's effect sweep is now URLSession-only network, no
telemetry, no database, keychain storage; its entry files drop from
eleven preview files to its one real script. SwiftPackageIndex's Fluent
database and reply steps are unchanged.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
Colby Mchenry
|
e9be68b61d |
fix(swift): a composition continued on & lines no longer swallows its type (#2108)
typealias EditorClient = AutocompleteService.Client
& MediaUploadService.Client
is valid Swift, but in a type's body tree-sitter-swift ends the member at
the newline and the `&` line is an ERROR that swallows the enclosing
type. IceCubesApp's 980-line `EditorStore` came out as loose variables
and functions — no class, no init, no methods — so a call in its init
counted as file-level code ("a file that runs something"), and calls
into its methods from the views landed on same-named methods of other
services.
A Swift preParse moves a line-leading `&` onto the line before, for an
indented `typealias`/`let`/`var` statement's continuation lines only.
Same length and line count; the continuation line's tokens sit one
column left (the Swift type gate reads the site's qualifier at that one
column over). Measured shapes: the failure is type bodies only — a
top-level declaration, a function body and a parenthesized type parse as
written, and are left alone; `&&`, `&+` and inout `&x` are never
touched. The preParse is hoisted before the kernel call like the C/C#
ones, so both arms parse the same bytes, and the file no longer defers
to wasm.
Across IceCubesApp, SwiftPackageIndex-Server, swift-nio, Alamofire, TCA
and mastodon-ios (3,000 Swift files) it rewrites exactly the one failing
file, which now parses clean; none newly erroring. EditorStore is a class
again with 38 methods; views now reference it and call its
insertStatusText / runAssistant (62 edges into it), and
`urlSession.upload` no longer lands on MediaUploadService.upload.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
Colby Mchenry
|
5753de20f5 |
fix(swift): an extension is not the type it extends (#2107)
`extension View { … }` parses as a class_declaration, so the index held a
`class View` node for every file extending SwiftUI's View, Text, Color,
String, URL. With no declaration of the type in the project, every
`struct X: View`, `Text("…")` and `Color.red` bound to whichever
extension came first: on IceCubesApp 6 of the top 12 "most depended on"
symbols were extension files (View 632 edges, Text 402, Label 209). And
a bare `@State` bound to a view model's nested `enum State` (253 edges),
`@Test` to a test's nested struct (821 on SwiftPackageIndex), `: Error`
to a nested `IOError.Error` (233 on swift-nio).
A gate at resolveOne's seam (every strategy, framework patterns and
name matching alike) checks a Swift reference's type target against what
the site wrote:
- an extension moves to the declaration it extends, or the ref stays
unresolved when the SDK declares the type;
- a bare name of a nested type needs the site inside its parent, the type
itself, or a subtype that declares inheriting from the parent;
- a qualified name must fit its path. The index keeps `Build.Id` as a ref
named `Id`, so the qualifier is read back from the source line
(generic arguments stripped: `EventLoopFuture<T>.Isolated`); a member
indexed top-level because an `#if` ended its type's body early still
fits a parent declared in its file.
Redirects pick the unique declaration in scope, then the same file, then
the nearest directory; a tie is left unresolved.
Two things rode on the wrong edges and are kept:
- `getSupertypes` named `View` for `struct X: View` only because that
edge resolved to an extension. Swift conformances to a type the
project only extends are now read from the declaration head (once per
node, memoized; a per-node unresolved_refs query was +75% index time on
swift-nio), so `X().withEnvironments()` still resolves to the
`extension View` method.
- interface-impl dispatch read a protocol's methods off the node the
conformance landed on. Swift protocol requirements are not extracted as
methods; the default implementations live on `extension P` nodes, which
are now included — every extension, not one arbitrary one (+359
dispatch edges on swift-nio, +44 on Alamofire, none lost).
Measured edge diffs vs main, every removal classified:
IceCubesApp 2473 removed (172 redirects to the declaration, 1989
SDK-extension edges, 280 wrong bare nested-type bindings, ~30 from the
class the typealias parse error swallows); SwiftPackageIndex 3810 (1461
redirects); swift-nio 10991 (5085 redirects); Alamofire 1368 (198). Index
time at parity (swift-nio 4.1s vs 4.0s, Alamofire 1.33 vs 1.37,
IceCubes 1.31 vs 1.31).
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
Colby Mchenry
|
48b1a19b19 |
fix(vapor): bind a route to the handler on the type it names (#2106)
`use: SearchController.show` kept only `show`, and name matching bound every controller's `show` route to whichever `show` came first — on SwiftPackageIndex-Server 7 of 13 routes pointed at another endpoint's code (Search, HealthCheck, ValidateSPIManifest and ReadyForSwift6's `show` all at AuthorController.show). The route's handler ref now keeps the type: `Type@method` (dotted for nested types, `API.PackageController@get`), `@method` for `self.x` and a bare `x`, and a `let todos = TodoController()` receiver's type. The Vapor resolver claims these and resolves on the owner's qualified name: nested types, handlers declared in `extension Type` in another file, and `extension API.PackageController` (whose node carries only the last segment, so its declaration line decides). `self.x` / bare `x` take the method of the type the route is written in. Of one type's overloads, the one declared to take a `Request` wins. Two types, or none, leave the ref unresolved rather than guess. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
26d6ccece5 |
test(mcp): hold the watcher's sync so staleness-banner edits stay pending (#2105)
`MCP staleness banner > lists pending files under "Pending sync"` flaked in the full suite (2 of 4 runs): status showed "0 modified" and no Pending sync section. The tests used `debounceMs: 4000` to keep an edit pending while they queried, but since #1397 a lone edit syncs after a 300ms quiet window whatever the configured debounce. codegraph_status awaits a freshly spawned worker to count changes before reading the pending set, and under load that outlasted the window, so the sync had already cleared the entry. A 500ms pause before the query fails the status, banner and #1968 path tests deterministically. Hold the watcher's sync open with a never-settling spy in those four setups instead; no assertion changes, and the test that checks the banner drops after a sync still runs a real one. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
85dd8b2a15 |
test(watcher): wait for a live FSEvents stream before the #770 restart edit (#2104)
The symlink retarget/restart test flaked inside the full suite with "waitFor timed out". A traced failing run showed the edit made right after unwatch()+watch() was never reported: fs.watch() returns before macOS registers the stream (libuv serves a process's watches from one FSEvents stream, rebuilt on its own thread after each watch/close), and an edit made in that gap is dropped, not delayed — so no timeout would have helped. Poke a probe file under the link until the watcher reports it, then make the edit under test once. Applied after the restart (the observed failure) and after the retarget (the same race with a 100-300ms gap). Under a simulated 300ms registration delay the old test fails 3/3 and the new one passes 3/3; a dead post-restart stream still fails it. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
179ac85bd9 |
fix(steps): draw how a server-rendered endpoint answers; bare PHP calls are functions (#2103)
Found testing `codegraph ui` on Spring and Laravel: a server-rendered
endpoint's Steps picture never said how it answers. petclinic's POST
/owners/new showed its database write and no reply; BookStack's book page
showed a 404 and not the page it renders.
Spring MVC answers by what a handler RETURNS, which is no call:
- A new tree-sitter reader, returnsInTree / returnsForFile (graph/
branch-guards.ts, exposed as SiteReader.returns), lists the returns a
definition makes itself — never one in a lambda, a local function or a
nested / anonymous class — and the string a same-class constant holds.
- steps.ts: for a mapped method of a class that renders views (not
@RestController, not @ResponseBody), a returned view name / constant is a
200 render, `"redirect:…"` a 302, `"forward:…"` a forward, each under the
condition it is returned in (effectLink takes a preset effect for a site
that is no call). Non-literal returns are not guessed at.
- effects.ts: `new ModelAndView(…)` (200, or 302 for "redirect:…") and
`new RedirectView(…)` (302) are replies.
Laravel's `view(…)` and `redirect(…)` were already reply rules, but never
reached them:
- name-matcher.ts: a PHP call written without a receiver is a function
call, so only a `function` is a candidate — BookStack's every
`return redirect(…)` bound to ApiDocsController::redirect, every
`view(…)` to a `$view` field (and, through the fuzzy fallback, to the
class `View`). PHP refs record the call expression's column, so the call
text at that column decides. koel alone loses 60 such wrong edges
(`basename()`, `auth()`, `value()`, …).
- effects.ts: the PHP reply rule never matched a chain — rules see the text
with argument lists stripped (`response->json`), the rule wanted `()->` —
so `response()->json(…)`, `redirect()->route(…)`, `redirect()->back()->
withErrors(…)` were never replies; `to_route` added; redirect chains are
302s. PHP's SPL programming-error exceptions (InvalidArgumentException,
LogicException, …) are no longer replies, and a namespaced exception is
read by its class name.
petclinic POST /owners/new: 200 `view owners/createOrUpdateOwnerForm` WHEN
result.hasErrors(), 302 `redirect:/owners/${…}` WHEN !result.hasErrors().
BookStack GET /books/{slug}: 404, 200 `view`, 302 `redirect` (the
InvalidArgumentException from a helper is gone). realworld's REST login is
unchanged.
Not in scope: PHP has no site reader (no branch-guard rules), so Laravel
replies carry no WHEN and no arguments — `abort(Response::HTTP_…)` and
`response()->json()` show without a status.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
Colby Mchenry
|
dcd79ad3c5 |
fix(ui-map): count depth in folders that fork, so a Maven project opens on its packages (#2102)
A Maven / Gradle project keeps every line of Java under src/main/java/org/<company>/<app>/, and those folders hold nothing but the next one. The Map cut one folder per level, so petclinic drew the whole program as one `src/main/java/org` box, and the deepest grouping option (4) stopped at `src/main/java/org/springframework` — no setting reached the packages seven levels down. A folder with exactly one subfolder and no file of its own (read from every indexed file, tests included, so ids don't move when tests are toggled) now joins the level below it. moduleIdFor takes the set as an optional fourth argument — without it nothing changes — and pickDefaultDepth counts the same levels. Collapsing never splits a group at the same depth (a pass-through folder has one child), so repos without such chains keep their grouping; a lone chain like koel's `app/Console` → `app/Console/Commands` is renamed to the folder that holds the files. The box label (the wire `label`, which nothing read) now elides a chain of three or more folders — `src/main/java/…/petclinic/owner` — and ModuleNode and the width calculation use it; the tooltip and side panel keep the full path. Everywhere else the label equals the id, so other maps look the same. petclinic's default map: 8 boxes (owner, vet, model, system, the package's root files, resources) instead of one; realworld opens on io/spring's api, application, core, graphql, infrastructure. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
0419a51636 |
fix(synthesis): link a Spring or Laravel test's request to the route it reaches (#2100)
Tests reach a controller by URL, never by a call — MockMvc's
`perform(post("/owners/new"))`, Laravel's `$this->postJson('api/me')` — so
every controller those suites exercise read as untested: the viewer's
"No test reaches this" and explore's `tests:` line both walk callers.
petclinic's processCreationForm, hit three times by OwnerControllerTests,
said no test reached it.
A new synthesis pass (tier-synthesizer section 4, registered after the
cross-tier pass, gated on java / kotlin / php) reads test files only:
- Spring: MockMvc builders (with MockMvcRequestBuilders imported),
`request(HttpMethod.X, …)`, the Kotlin `mockMvc.get(…)` DSL,
WebTestClient `.get().uri(…)`, TestRestTemplate, RestAssured.
- Laravel: `$this->get/post/…Json(…)`, `->json('POST', …)`, `->call(…)`
on a chain rooted at `$this` that only calls methods (so
`$this->app->get('config')` is not a request); Pest's helpers where the
file imports Pest\Laravel; and a project's own verb-named helpers
(koel's getAs → jsonAs → `$this->json($method, $uri)`), recognised by
their body, never by name alone.
Paths are read as built — Spring `{var}` templates, PHP / Kotlin
interpolation and concatenation become holes — and matched with the
cross-tier pass's own route matcher (a hole fills a parameter, never a
literal). One `calls` edge per (test, route), `synthesizedBy:
'test-request'`. These edges are excluded from countIncomingEdges, so a
well-tested endpoint never becomes a hub the Steps walk refuses to enter.
Routes reached by a test: petclinic 16/17, realworld 18/19, BookStack
194/340, koel 105/134 (from 0 each). A random sample of edges checked
against the test source was all correct. The pass costs 86 ms on
BookStack and 68 ms on koel.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
Colby Mchenry
|
ffced05e3c |
fix(laravel): name routes by the path a request takes — leading slash, prefix groups, the /api mount (#2099)
Routes were named by the bare literal in the routes file, so BookStack's
api.php showed `GET pages` for `/api/pages` and koel's `Route::prefix('api')`
group showed `GET ping` for `/api/ping`. Beyond the label, a front-end
`fetch('/api/pages')` could never pair with a route recorded as `pages`.
- extract(): `Route::prefix()` chains and `Route::group(['prefix' => …])`
scope their routes, nested ones in order; a computed prefix is kept as a
`{prefix}` segment. Paths are joined the way Laravel joins them (leading
slash, no doubled or trailing slashes). qualifiedName now holds that
in-file path.
- postExtract(): finds where each routes file is mounted — a group whose
argument is the file, a require / include / loadRoutesFrom inside a
group, Laravel 11's `withRouting(api:, web:, apiPrefix:)` — composes the
prefixes down the include tree, and names every Laravel route from its
qualifiedName plus its file's prefix (idempotent; a removed mount is
undone). routes/api.php defaults to `/api` unless something mounts it or
a `withRouting(using:)` loader might. A file mounted at two prefixes stays
at its in-file path. Only files in routes/ or naming `routes/` /
`withRouting(` are read.
On BookStack all 80 api.php routes become `/api/…` (from its
RouteServiceProvider group); on koel all 118 become `/api/…`; no route is
left without a leading slash.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
Colby Mchenry
|
2b9f2f47ef |
chore(ui): version the component package with the engine (1.6.1) (#2098)
The 1.6.1 bump changed package.json only, as AGENTS.md said was enough, so
ui-package.test.ts ("is versioned with the engine") has failed on main since.
Bump ui/package.json to match, and say in AGENTS.md that a version bump
touches both files.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
github-actions[bot]
|
f4ddf50851 |
docs(changelog): promote [Unreleased] into [1.6.1]
[skip ci] Auto-generated by Release workflow.v1.6.1 |
||
|
github-actions[bot]
|
0695bc47ce |
release: sync package-lock.json to 1.6.1
[skip ci] Auto-generated by Release workflow. |
||
|
Colby Mchenry
|
4a8eb6314e |
release: 1.6.1 (#2090)
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
2c4d4eb9e4 |
release: withhold codegraph ui until the viewer launches; hold its changelog entries (#2089)
* feat(cli): withhold `codegraph ui` until the viewer launches (CODEGRAPH_UI=1 opts in) The viewer has been exercised on Expo apps only, not on API, Laravel, Spring, Angular or Swift projects, so it stays out of this release. `ui` and its alias `web` — also as `help ui` or `ui --help` — are refused before any startup work, and the command is hidden from `--help`, unless CODEGRAPH_UI=1 is set, which keeps it usable for the people testing it. The viewer assets still ship, so the opt-in works on a released build. Reverting this commit is the launch. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(site): say the viewer is not in a release yet The viewer pages have been live on the docs site since they landed, and 1.6.0 answers `codegraph ui` with "unknown command" (#1816, #1900). Until the viewer launches, the guide, the CLI reference and the next-steps link say so. Reverting this commit is part of the launch. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(changelog): hold the viewer's entries for its launch; new Highlights for the fixes release 53 entries that describe `codegraph ui` and its screens move, verbatim, to docs/viewer-launch-changelog.md along with the launch Highlights. 12 entries whose graph and `codegraph_explore` half ships now (the router navigation edges, the cross-tier hops, explore's `when` conditions, SvelteKit layouts, wrapped Express handlers, …) are reworded around the graph; their original wording is kept in the holding file for the launch. A few kept entries lose a viewer clause. The Highlights are rewritten for this release. prepare-release.mjs promotes the section and extract-release-notes.mjs reads it back with no mention of the viewer; the released sections are untouched. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
09b9646b7e |
perf(context): compute the dominant file once per index state (#1864) (#2086)
* perf(context): compute the dominant file once per index state (#1864) getDominantFile() backs context ranking's core-directory boost. Its answer depends only on the graph, but the whole-graph aggregation behind it (edges joined to both endpoints' nodes, grouped by file) ran on every generic codegraph_explore - seconds per call on a large index. Memoize it in QueryBuilder against a database change stamp: total_changes() (rows this connection wrote) plus PRAGMA data_version (commits by any other connection or process). Both are O(1), so no write path has to remember to invalidate, and a sync made by another process is seen on the next call. Any write forces a recompute; the heuristic's result is unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(context): never keep a dominant file read inside a transaction (#1864) total_changes() counts writes that are later rolled back, so a result computed inside a transaction could survive the ROLLBACK under an unchanged stamp and describe edges that no longer exist. getDominantFile now skips the memo while a transaction is open (and on a runtime without isTransaction). Ported from the maintainer's hardening of this fix in #1919. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(db): drop the dominant-file memo when a QueryBuilder is rebound (#1864) The memo's change stamp (total_changes + data_version) is per connection, and fresh read-only connections to two different databases report the same one ("0:2"). A pool worker rebound onto a rebuilt index kept serving the old database's dominant file until something wrote through the new connection. A test rebinds onto a second project and fails without the reset. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(changelog): credit @danusha2345 for #1864 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: danusha2345 <danusha2345@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
b44550e328 |
docs(changelog): credit @danusha2345 for the #1902 fix (#2085)
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
ee97a57a9b |
fix(directory): a schema-less codegraph.db does not make a project initialized (#1895) (#2083)
* fix(directory): a schema-less codegraph.db does not make a project initialized (#1895) isInitialized() accepted any existing .codegraph/codegraph.db, so an empty or table-less file in an ancestor directory (an interrupted init, a stray touch, a never-populated ~/.codegraph/) captured the upward walk of findNearestCodeGraphRoot / resolveProjectPath / resolveServerRoot for every project beneath it: the real index of a sub-project became unreachable, the sub-project down-scan never ran, and commands either operated on the broken file or failed opening it instead of showing the not-initialized guidance. An initialized project is now one whose db carries the schema: the probe is gated by file size (>= the 100-byte SQLite header), then the header magic, then a read-only node:sqlite open with a single sqlite_master lookup for the `nodes` table, memoized per path + mtime + size so hot callers (the prompt hook, MCP root resolution on every call) never reopen an unchanged db. The handle is closed immediately so the probe never holds the db on Windows; a locked/busy db counts as live. createDirectory() uses the same predicate, so `codegraph init` in the directory with the broken file repairs it in place (the schema is applied with CREATE ... IF NOT EXISTS) instead of refusing with "Already initialized"; the CLI says what it found. Nothing is deleted automatically. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> * fix(directory): a database we cannot open read-only still counts as initialized The schema probe failed CLOSED: a real WAL codegraph.db in a directory the process cannot write (a read-only checkout, a mount, another user's tree) made the read-only node:sqlite open throw — SQLite must create `-shm` to open a WAL database — and isInitialized() reported the project as not initialized, where main's file-exists check had been right. The probe now fails OPEN. Once the size gate and the `SQLite format 3\0` header have passed, the file is a SQLite database; only a SUCCESSFUL sqlite_master query proving the `nodes` table absent may return false. Any open/prepare error (locked, busy, cantopen, readonly, disk I/O) returns true, the pre-existing behaviour for a database we cannot inspect. No error strings are matched. The per path + mtime + size memo is unchanged, and schema-less/empty files still return false, so `codegraph init` still repairs them and never touches a real db. Test: a WAL db copied into a `.codegraph/` made read-only with chmod stays initialized — skipped on win32 and when running as root (root ignores mode bits). Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> * fix(init): refuse a codegraph.db that is not SQLite instead of promising a rebuild Review found that a `.codegraph/codegraph.db` holding bytes that are not SQLite at all counted as "schema-less", so `init` printed "rebuilding it" and then failed with SQLite's "file is not a database". Only an empty file or a SQLite database without the codegraph tables can be rebuilt in place; `hasSchemalessDb` now means exactly that, and the new `hasForeignDbFile` names the other case. `init` stops on it with exit 1, says which file it is and to move or delete it, and never touches it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(directory): ask SQLite what codegraph.db is instead of reading its header ourselves (#1895) The header check opened and closed a descriptor of our own on `codegraph.db`. Closing any descriptor on a database file drops every POSIX lock the process holds on it, including those of a connection it already has open (sqlite.org/howtocorrupt.html §2.2.1) — and the MCP server resolves projects through isInitialized on every call while it holds the index as its writer. mcp-projectpath-lifecycle's takeover tests caught it: after a real owner exits, the taking-over engine's next read failed with "disk I/O error" (2 of 21 fail with the header read; 21/21 on main and with this change). One read-only SQLite probe now answers all three questions: the schema is there, SQLite opened it and the schema is not (repairable), or SQLite refused it as not a database (SQLITE_NOTADB). Anything else — locked, busy, no `-shm` possible — still fails open as initialized. SQLite's own connections share one lock table per file, so opening and closing a second one leaves the first one's locks alone. A test pins that isInitialized and both init helpers never open the file themselves while this process holds it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(changelog): keep the #1895 entry clear of #1917's, so the carries merge in any order Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: danusha2345 <danusha2345@users.noreply.github.com> Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
04ec5f554d |
fix(extraction): never read an oversize file; an MPEG-TS video named .ts is not TypeScript (#1910) (#2082)
* fix(extraction): an MPEG-TS video named .ts is not TypeScript (#1910) `.ts` mapped to TypeScript by extension alone, so MPEG transport-stream fixtures (`testdata/*.ts`) were parsed by tree-sitter: ~28 s of CPU for a 900 KB clip, minutes for a directory of them, for zero symbols. Recognise the stream from the head of the file — the 0x47 sync byte at offsets 0, 188, 376 and 564 plus a NUL byte, which every stream carries in its first packets and UTF-8 source never does — and drop it at discovery: not indexed, not parsed, not counted, not tallied as an unsupported language. The check reads under 1 KB and only for `.ts` files; the batch reader sniffs the bytes it already read, and the single-file path (sync, watcher) does the same head read, so a clip handed in by name is skipped too. The skip happens before parse dispatch, so no kernel mirror is needed. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> * fix(extraction): a video .ts is never pending on the git path or a scoped sync Review found two places that decide "is this a source file" without the MPEG-TS check the scan applies: the git fast path's candidate loop in getChangedFiles and the scoped-sync path filter (watcher events). An untracked clip in a git repo was reported as `added`, skipped by sync without a record, and reported as `added` again on every status; a tracked .ts that became a clip stayed `modified` with its stale nodes. Both now treat the clip as non-source: removed when tracked, ignored otherwise. Regression tests fail without this change and pass with it, on both arms. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(extraction): require a binary head before calling a .ts file video (#1910) Review on #1915: the sniff took four aligned 0x47 sync bytes plus any NUL as proof of a transport stream, so TypeScript with a `G` at four 188-byte strides and one raw NUL in a comment was dropped unindexed. The sniff now needs the sync byte on sixteen consecutive packets (3 KB of head) and at least 1/64 of the head to be control bytes. Compressed payload carries 8-20% of them (checked on ffmpeg-made H.264/AAC, MPEG-2 and MP2 streams); source text carries none. A shorter clip is cheap to parse anyway. Also drops the benchmark figures from the changelog entry. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(extraction): a file over the size limit is never read (#1910) The size gate ran after `readFile`, so a committed video or blob fixture was decoded in full — 3.4 GB of RSS for a 400 MB file, `Invalid string length` past ~512 MB — only to be stored as skipped. It was then read again by every pass that scans files by name: the framework detectors' `readFile` and the resolver's cached reader, three full passes on the reporter's fixtures. Stat first, everywhere a source file is read by path: the batch reader and `extractFile` store an oversize file with a size stamp in place of its content, change detection hashes the same stamp (so a same-size rewrite of a file nothing is indexed from is not a change, while crossing the limit in either direction is), and both by-name readers return null above the limit. The limit moves to `src/file-limits.ts` so the three sites share one number. strace on a sparse 400 MB `.ts`: 153 603 reads before, 0 after; indexing it next to a real file costs no measurable RSS. Stacked on #1914, which shares the batch-reader hunk. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> * refactor(extraction): one-purpose size-stamp reader, comments in operation order `readSourceOrStamp` took a `stats` argument no caller passed and returned stats no caller read; it now returns the text or stamp change detection hashes. The batch reader's two comments are back in the order its code runs: the size gate, then the MPEG-TS head check. No behaviour change. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(extraction): bound the read itself, not only the stat before it (#1910) A file can grow between the stat that gates the size limit and the read that follows (a log, a download, a build output being written). Every source read now goes through readBoundedSource/readBoundedSourceSync: the open descriptor is re-checked, and the read stops one byte past the limit, so an oversize file is still stamped rather than decoded. The extraction batch reader, single-file extraction, framework detectors and the resolver's file cache all use it. Ported from the maintainer's hardening of this fix in #1919. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(extraction): bound the parse-retry reads too (#1910) The retry passes for files whose worker crashed or timed out re-read the source with an unbounded fsp.readFile. The file can have grown since the first, bounded read; an oversize file is now skipped by the retry instead of being decoded. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(extraction): check an oversize file against its size stamp everywhere (#1910) Review on #1915: - The viewer (readFileShape, hasDriftedOnDisk, the source endpoint) and MCP's drift gate hashed a file's real bytes, but the index stores the size stamp for a file over the limit, so every unchanged 1-8 MB file read as "changed on disk". oversizeStamp moves to file-limits.ts with an indexedHashInput helper, and all four checks hash what the index stored. An oversize file is now compared without being read. - git-index-currency's "keeps a committed path pending when sync cannot read it" injected its failure only into readFileSync, which the bounded reader no longer calls, so it failed. The failure now reaches openSync too. - The changelog entry drops its benchmark figures. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(mcp): check an oversize answer file against its size stamp (#1910) #2043's answer-freshness check hashes each contributing file whole and compares it with the stored contentHash. For a file over the index's size limit the stored hash is the size stamp, so an unchanged oversize file would read as stale. Compare it by its stamp, without reading it, like the viewer and the MCP drift gate do. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * perf(extraction): judge a .ts clip from the bytes it is read for, not at discovery (#1910) Discovery sniffed the head of every `.ts` file on every full scan — an open + 3 KB read + close per file, before anything asked for its content. On a 10k-file TypeScript tree that took a warm-cache scan from ~105 ms to ~243 ms; on a slow disk or a network share it is a random read per file per scan, the access pattern #1231 and the SMB reports (#1014, #448) are about. The bytes are now judged where they are read anyway, so the check costs no I/O and runs only for a file that is new or changed: the batch reader (as before), `indexFile`, the sync reconcile (an untracked clip is skipped, a tracked file that became one is removed through the same removal path as a deletion), and both `getChangedFiles` paths. The scan lists `.ts` files by name again; a test pins that discovery opens none of them on the walk or the git path. Scan time is back to main's (~98 ms), and a full index is unchanged within noise. Also collapses the four #1910 changelog bullets (#1914 and #1915 each added both) to two, without the benchmark figures. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: danusha2345 <danusha2345@users.noreply.github.com> Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com> Co-authored-by: danusha2345 <ewidusoc498@gmail.com> |
||
|
Daniil
|
e5a3d85b0c |
fix(watcher): follow a rebuilt index instead of syncing into the replaced file (#1902) (#1917)
* fix(watcher): follow a rebuilt index instead of syncing into the replaced file (#1902) `codegraph index` rebuilds through `CodeGraph.recreate()`, which unlinks `.codegraph/codegraph.db` and creates a new file at the same path. A running MCP daemon kept its handle on the unlinked inode and its watcher kept "auto-syncing" into it, so no edit reached the on-disk index. The #925 self-heal only ran on the tool-call path and ran no catch-up after reopening. - `sync()` now checks, once it holds the index mutex and the write lock, whether the database was replaced on disk (one stat, the same inode check `reopenIfReplaced` uses). If so it reopens the live file in place and widens a scoped sync to a full reconcile. A failed reopen (rebuild mid-way) returns the lock-busy shape so the watcher keeps its pending files and retries. - `reopenIfReplaced()` refuses while an index/sync holds the mutex, so the tool-call path never closes the handle an in-flight sync writes through. - When a tool call's `freshen()` does reopen, the engine runs its existing catch-up sync (serialized on the same mutex), but only for its own watched instance, so an engine that is not the project's writer never writes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(sync): step aside for a rebuild that has recreated the file but not locked it yet `codegraph index` recreates the database, then takes the write lock inside indexAll. With the watcher now following a replaced file, a sync landing in that gap reopened the empty file and started a full reconcile, holding the lock the rebuild was about to request — `codegraph index` could then fail with "another process may be indexing". Review flagged the gap; this closes it. After following a replaced file, a sync that finds no index_state yet (the rebuild has not started indexing) and a file written within the last two minutes returns the lock-busy shape so the watcher retries; the first sync after that reconciles the whole tree once, then syncs go back to scoped. The time bound keeps a rebuild that died before indexing from parking the watcher forever. New test: fails without the guard, passes with it; 136 related tests pass. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(sync): keep the full catch-up until a full sync completes (#1902) Two gaps in following a replaced database: - Only a reopen done by sync() itself set pendingFullReconcile. The tool-call self-heal (reopenIfReplaced) swapped the handle without it, so the next scoped watcher sync skipped whatever the old handle had absorbed. The flag is now set wherever the database is reopened. - The flag was cleared before the sync ran, so a sync that threw lost the catch-up. It is now cleared only after a full sync completes. Ported from the maintainer's hardening of this fix in #1919. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(sync): report a rebuild in progress as lock contention, not an empty sync (#1902) Since #2014 the watcher tells lock contention apart from success only by LockUnavailableError; it no longer recognises the all-zero result. Following a rebuilt index still returned that shape when the reopen failed or when the recreated file had not been indexed yet, so the watcher counted the step-aside as a clean sync and dropped the pending edits. Both paths now throw LockUnavailableError. A watcher test proves the edit stays pending through the rebuild gap and is reconciled once the rebuild finishes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: danusha2345 <danusha2345@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: danusha2345 <ewidusoc498@gmail.com> |
||
|
Colby Mchenry
|
15a5d1d0cd |
fix(c): keep an #undef under an undecidable #if unknown; a macro call never binds to a type (#2069, #2070) (#2081)
Two follow-ups to the unreleased macro-visibility work (#1838/#1839): - #2069: `#undef FLAG` under an `#if` nothing in the source decides made a never-seen FLAG certainly undefined, so `#ifdef FLAG` was read as false and the functions it guards lost their callers (CPython's `Py_GIL_DISABLED`). A name no directive has touched is now unknown; only an #undef on a certain line clears it. - #2070: dropping the macro constant from a call's candidates let the next same-named node win, so expat's `PREFIX(scanRef)(...)` instantiated an unrelated `struct PREFIX`. When the name is a function-like macro in the project, only a function or method is an eligible callee. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
fef37765c1 |
perf(resolution): remove the indexing regressions since v1.6.0 (vscode stall, per-ref rescans) — graph byte-identical (#2072)
* perf(resolution): read JS/TS parameter lists without exponential backtracking isLocallyBoundJsName (#1759, |
||
|
Colby Mchenry
|
c3b1348065 |
fix(indexing): reject indexAll after a failed store instead of spinning (#1773) (#2079)
The in-order commit window waits for the commit cursor to advance before dispatching more parses. After a store fails, flushOrdered returns at once and the cursor never moves, so the backpressure loop spun on microtasks forever: indexAll never settled, a core stayed pinned, and no timer in the process could fire. It reproduces whenever a store error (a closed database, disk full, an I/O error) arrives with more files left than the window holds. In the test suite, which runs from source with a window of one, a test that timed out mid-index and closed its graph in afterEach wedged the whole worker, so a loaded Windows run never produced a report. Stop waiting once a store has failed or the run was aborted, and destroy the parse pool on that error path so its worker threads do not outlive the index. The regression test runs the scenario in a child process under a hard timeout. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
c405ac5e6c |
fix(windows): make load-sensitive tests wait on real signals; longer pid-file retry (#1773) (#2078)
* test: make Windows load-sensitive tests wait on the tree, not the clock (#1773) Each failure from the loaded Windows run was traced to its mechanism and reproduced deterministically before changing it. No assertion is weakened. - daemon-lock-refresh: the PowerShell holder slept 200ms against a 375ms retry budget. Windows refuses a replace-rename while any handle to the target is open, even a Node read handle, so hold one in-process and close it inside the rename mock after the first real failure. Also asserts the first rename really failed with a retryable code. - mcp-projectpath-lifecycle: the owner daemon armed a 500ms idle timer at start and exited (code 0) before a starved engine reached it. It now starts with idle exit off and is armed over IPC once the engines hold a session. The file waits for catch-up instead of the gate's 3s serve-anyway deadline, and its index-heavy hooks and eviction case get explicit timeouts. - mcp-daemon / mcp-writer-lock teardown: racing launchers spawn detached daemon candidates whose cwd is the fixture; a loser still starting at teardown pinned it (EBUSY) or took over after the winner was stopped. A --require preload records candidate pids, and teardown waits for the losers before stopping the winner. - mcp-writer-lock: writer.pid appears before the daemon binds and rewrites daemon.pid, so cleanup could probe a daemon mid-start ('unverified'); wait for both proxies to attach. Removal uses fs.promises.rm, because fs.rmSync gives up on the first Windows access-denied error despite maxRetries. - mcp-daemon #1963: connect only after the proxy attached (lock before bind). - sync-rebuild-convergence / multi-repo-workspace: explicit timeouts scoped to the cases that build real indexes; the global timeout is unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(daemon): give the Windows pid-file replace a longer retry budget (#1773) Windows refuses to replace a file while any handle to it is open, including a plain Node read handle, so an antivirus scan, an indexer, or another session reading daemon.pid can each block the daemon's startup refresh for a moment, and for longer on a loaded machine. The retry waited 375ms in all; it now backs off 25, 50, 100, 200, then 400ms three times, about 1.6s over eight attempts, still well inside a launcher's ~6s connect window. Permanent failures and non-Windows errors still abort at once, and an ownership change still stops the retry. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
003305f1e7 |
fix(explore): claim complete source only for sections that are complete (#2077)
* fix(explore): claim complete source only for sections that are complete On the tiers with a completeness signal (>= 500 indexed files) every codegraph_explore response ended with "Complete source for N files is included above — do NOT re-read them", whatever the render had cut. A windowed spine method, a shrunk or dropped cluster and the per-symbol view all went out under that line; the oversize-spine window never even set the trim flag the small tiers key off, which is how a 62-line slice of vscode's 968-line rpcProtocol.ts was called complete. The line also said "Reserve Read for a single specific line range", which explore output must never say. Completeness is now measured at emission from the ranges actually sent (plus back-referenced spans) against the symbols each section set out to deliver. A trimmed response keeps the verbatim / already-Read guarantee, names the trimmed files and the most relevant elided symbols, and steers to another codegraph_explore. The note is offered from most to least specific and fitted to the leftover room (CG-26); a trimmed note yields the pointer list's first entry. The lost-pointer, cut and truncation notes drop "complete" the same way, at no greater length, so the epilogue floor is unchanged. The small tiers' trimmed note now also fires for the cuts the flag missed. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(explore): offer an elided method as Owner.member in the completeness note The note offered `as_sql` for django's elided SQLCompiler.as_sql — one of 110 definitions — so agents appended a path and line to disambiguate it, and that query shape returns none of the body. `SQLCompiler.as_sql` alone returns most of it. Methods are now offered as `Owner.member` from the indexed qualified name; everything else keeps its bare name. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * refactor(explore): fit the completeness note and pointer list in one tested function A trimmed note's optional detail (the files, the elided names) never costs a pointer entry the least specific note would have left. The fit moves out of handleExplore into fitExploreEpilogue so both rules are pinned by tests; the second-try branch it had could never fire and is gone. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * test(explore): completeness note wording after #2068 flags the spine window Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(explore): record the completeness-note replay and agent A/B Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(explore): address review of the completeness note - Judge completeness AFTER the dedup fold: a remainder too small to fence is sent in neither call, so a section with one is trimmed, not complete. The whole-file render now declares its symbols too (clamped to the lines it prints; the file node is not one), so a fold there registers. - Label trimmed files unique among every path the response can name, the pointer list included. - "1 file", and no "0 files" when every section was held from an earlier call. - The lost-pointer, cut and truncation notes are one exported table, EXPLORE_FALLBACK_NOTES, with the length rule tested; the lost note picks its wording from trimmedShown instead of re-joining the response. - The design doc's follow-up list marks the overclaim fixed. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
290e03f7bc |
fix(explore): pin only the same-named file that defines the symbols a query names (#2071)
A bare basename several directories share (`editorOptions.ts`, `worker.rs`, `compiler.py`) pinned every match, and the matches split the pinned reservation: on vscode `editorOptions.ts clampedInt cursorStyleToString cursorStyleFromString` gave the file holding all three 8,626 chars, against 15,181 without the path, and spent the rest on a workbench helper defining none of them. Path extraction now takes an injected `symbolFiles` lookup (the module stays DB-free, like `existsOnDisk`). When a span matches several files, the matches defining a precise symbol token the query names (camelCase, PascalCase, snake_case, qualified — the seeder's NL-stopword shape test) are pinned and the rest set aside; the summary line names them. When no match defines one, or every match does, the span resolves exactly as before. The same test resolves a basename shared past the ambiguity budget (django's `models.py`) when it narrows into budget, and a line anchor binds once the span narrows to one file. Replay of 196 queries over 8 indexed repos, main vs this branch: every unambiguous pin is byte-identical (38 unique basenames, 13 dir/basename, 8 file:line, 38 earlier-sweep path queries), as are 26 path-free queries, 13 where every match defines a named symbol and 8 naming no symbol. All 52 changed responses are shared basenames — named FULL 207 -> 231 (11 up, 0 down), named lines 6,045 -> 7,577, none over 25K. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
724b5dee0c |
fix(explore): return named bodies from merged spine clusters; let named files use spare budget (#2068)
* fix(explore): return named bodies from merged spine clusters; let named files use spare budget Re-applied onto main after #2062 and #2063 landed in the same render path (originally e31cc9fc on claude/inspiring-grothendieck-8a7b3e): - the merged-spine hold-back and its top-up are dropped: #2062's hold-back (every cluster leaves room for the protected members ranked below it) subsumes them, and the past-the-pin-cap test passes on it; - #2063's "don't window a cluster holding an exact target" becomes the member rule's "don't window a spine method that is or holds one"; - the shrink orders exact targets ahead of spine members, and the exact branch prices member windows (sectionRangesOf) rather than whole bodies; - the final ceiling fit shortens the file header back to its estimate before it cuts source. With an exact target's overshoot (#2063) spending a named file to its funded line, a lower file's header running past its estimate cost it 109 chars of source (922 of 1,031; 995 on main). The valve test now asserts funding in full and delivery to the line, since #2062's exact rendering stops at whole lines. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(explore): record the agent A/B for the named-concentration fix Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(explore): record landing named concentration on main after #2062/#2063 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
7506dbff69 |
test(explore): guard named symbols in a pinned file past the node cap (#2064)
A file named by path is pinned: its first 300 symbols by start line enter the gather. Past that cap the file splits into a head cluster of pinned filler and, one cluster down, the named symbols the cap cut off, which used to render nothing (`protocol.ts serializeAck`: 0 of 3 lines; the same query unpinned: all of them). #2062's cross-cluster hold-back fixes this shape without knowing about pinning; its own fixture is an unpinned file. This fixture pins the pinned entry path to it: red on tools.ts before #2062 (2 of 6), green on main. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
47840a6ed4 |
fix(explore): return a qualified or line-anchored method's body instead of its signature (#2063)
`SQLCompiler.as_sql pre_sql_setup get_select` on django returned get_select and pre_sql_setup in full and SQLCompiler.as_sql -- qualified, step 1 of the Flow, 226 lines -- as one signature line. compiler.py renders as a per-symbol focused view, which picked bodies in source order within a tier, so the unnamed bridge get_qualify_sql and get_select (both above as_sql in the file) took the cap. The follow-ups agents issue (`compiler.py:776`, `compiler.py lines 900-1003`) pinned the file and dropped the line numbers. Every agent run ended in a Read of compiler.py. - query-paths: keep `:776` / `:12-40` / `#L88-L120` suffixes as line anchors and bind prose ranges (`lines 900-1003`, `L900-L1003`, `lines 900 to 1003`) to the nearest single-file path. - EXACT targets: a qualified name resolving to <=3 callables, or the callable enclosing a single-line anchor. Multi-line anchors render as that span. - Focused view: tiers exact > named spine step > unnamed bridge > unique named > family co-named, spine in call order. A tier 0-1 body that does not fit is windowed (head + calls into the question's other symbols) with the explore range that returns each hole, never reduced to a signature. Priced in rendered chars net of the signature lines, so the section keeps its promise. - Cluster path: exact clusters rank first with the spine's bounded overshoot, shrink by exact rendered size, and hold back what lower-ranked exact clusters still owe; files holding one are funded like spine files. - Blast radius leads with exact targets. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
adad8407ff |
fix(explore): pay named members before incidental ones across clusters (#2062)
Clusters of equal max importance rank by density, and density counts every member, so a named function with unrelated helpers merged around it outranked a cluster holding a named function alone. Taken whole, the higher clusters spent the file's reservation on the helpers and the isolated named function rendered nothing (lib/response.ts repro: sendBody 0 of 36 lines). Reordering alone only moves the victim. Two rules, both scoped to incidental members (not named, not an entry point, not on the call path): - a cluster's incidental members may only use what is left once every lower-ranked cluster's protected members are paid; its own protected members are never held back, so named-vs-named stays rank's call; - in a cluster holding a protected member, incidental members are never kept past the render's ceiling in rendered size, so the source-order window back to the ceiling no longer cuts a named body's tail (gin's handleHTTPRequest) on selection or on any re-render. 47 deterministic queries over the 7 README repos plus express vs main: named bodies complete 198 -> 234 (17 responses up, 0 down), named lines +10%, responses over the 25K cap 5 -> 1. One new final-cut truncation (django-bag) comes from owedPayableBelow's MIN_CHARS threshold, recorded in the design doc. On main + the named-concentration branch: 235 -> 279. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
c13bd0890f |
fix(wsl): keep WSL off Windows' index on /mnt drives, and explain the I/O error (#995) (#2061)
When Windows-native and WSL CodeGraph both open the same .codegraph/codegraph.db under /mnt/<drive>/, SQLite's locking and -shm shared memory don't hold across the 9p/DrvFs bridge and WSL fails with a bare "disk I/O error". - A SQLite I/O error on the default index of a WSL /mnt/<drive> project is rewritten (in the SQLite adapter, so every open and query is covered) into the CODEGRAPH_DIR=.codegraph-wsl instruction. MCP answers it success-shaped, and a failed default-project open is reported instead of "no project loaded". - On WSL + /mnt/<drive> with CODEGRAPH_DIR unset, the data dir is resolved per root: an existing .codegraph-wsl/ wins, then an existing .codegraph/ index (kept, no silent re-index), else a fresh index goes in .codegraph-wsl/. `codegraph init` says where it went. Every other host is unchanged. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
e63fe2ec5b |
test(wal-heal): await the heal open() started instead of a wall clock (#2056)
`open() itself kicks off the heal` polled the WAL size for 30s and then asserted it. On a loaded machine the heal was still running when the deadline passed, so a slow heal failed as a size assertion, reading like a regression (#1773). Await the pass open() started instead: it still fails at once if open() stops kicking off the heal, and reports the heal's own result rather than the clock's. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
66aebd17e2 |
fix(explore): three retrieval regressions found against v1.6.0 (#2057)
The pre-release agent A/B (main vs v1.6.0, 7 README repos, 3 runs each) had main returning 15% less source on vscode for the same queries, with Reads in 2 of 3 runs where v1.6.0 had none. Replaying every query both builds' agents issued traced it to three causes: 1. Named gaps (#1711) displaced source. Every cluster fit measured the gap names, which on a long path run to hundreds of chars, so a class shrunk into its file's room overran it by the names alone and was dropped whole: rpcProtocol.ts emitted 222 of 5,496 funded chars (v1.6.0: 5,194), and the test fixture came back as an empty fence. Fits now measure bare gaps; names are added at assembly from what the file's budget has left. 2. Interface members (#1638) corroborated English words. The NL-stopword guard lets a bare word seed a definition when another query word names a symbol in the same file; `readonly host` in an options interface made "main" seed `main()` in agentHostServerMain.ts, which took the named-first tier from the answer files. Members declared inside an interface no longer count as corroboration. 3. The interface-dispatch note announced dispatch through any common base, declared or not: "`extension` -> runtime dispatch to 2706 types implementing Disposable" opened 9 of 31 vscode answers. A supertype now has to declare the member (it, an ancestor, or a same-named Swift extension). When absence can't be judged, because a protocol's requirements aren't indexed, the note stays as before. Replayed queries, source lines vs v1.6.0 (main -> fixed): vscode 10,744 -> 12,577 (12,617), excalidraw 13,065 -> 13,698 (13,764), django 8,420 -> 8,803 (8,954), tokio 13,618 -> 14,151 (14,600), okhttp 6,511 -> 6,883 (7,090), alamofire 7,019 -> 7,249 (7,120); no query below 60% of v1.6.0. Real dispatch notes (Alamofire adapt/retry/asURLRequest, django, okhttp) are unchanged. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
49607201d6 |
fix(resolution): bind implements to the interface of a value+interface pair (#2055)
The inheritance target-kind gate (#1796) rejected any extends/implements resolution whose target cannot be a supertype. VS Code declares every service twice under one name, `export const IFoo = createDecorator<IFoo>(…)` beside `export interface IFoo`, and the import resolver takes the first export of that name, which is the value. The gate then dropped the edge instead of taking the interface the same file declares. Found by the pre-release v1.6.0 vs main comparison: 864 `implements` edges lost on vscode (AccessibilityService -> IAccessibilityService and ~860 like it), plus the ~5.8k interface-dispatch call edges synthesized from them. The gate now moves a TypeScript constant/variable target to the one same-named supertype in its file; with no such sibling it still drops the edge. vscode: implements 4,139 -> 5,003 (v1.6.0: 5,436, 254 of them on non-type targets; now 0), node count unchanged. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
1cbac0afaa |
docs(agents): update the Windows test baseline after #2053 (#2054)
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
0a9f3de72c |
fix: make the full test suite pass on Windows (#2053)
* test: close resolution graphs before removing Windows fixtures
Close ArkTS and JVM graph connections in afterEach, including failed assertions. Verified both files on Windows 11 and macOS (32 tests).
* test: compare explore source lines across LF and CRLF checkouts
Normalize only line endings when matching emitted source against checked-out fixtures. Preserve the closure coverage and whole-line assertions. Both files pass on Windows 11 and macOS (15 tests).
* test: exercise old Git fallback without a POSIX shell shim
Inject only the unsupported ls-files invocation and run every other Git command against real repositories. Assert that each rejected call retries, use portable graph paths, and clean up fixtures. Passes on Windows 11 and macOS.
* test: normalize Python fixture before generating CRLF cases
The tests added in #2004 produced CRCRLF from a Windows checkout. Normalize the input to LF before constructing either variant. All 28 native and WASM cases pass on Windows 11 and macOS.
* test: allow time for eviction fixture indexes on Windows
The #2036 eviction test builds and reconciles nine real indexes and consistently exceeds its default five-second timeout on the VM. Allow 15 seconds for this test alone, preserving every assertion. The full file passes on Windows 11 and macOS (21 tests).
* test: finish MCP processes before resetting Windows fixtures
Launch actual servers with their runtime flags and await termination before directory removal. Use asynchronous removal retries for daemon fixtures. Establish a ready daemon and stop its first proxy before simulating PID reuse, then make source stale without opening a competing SQLite writer. Preserve lock and database byte-equality assertions. All touched files validated on Windows 11 and macOS.
* test: allow Windows named-pipe fallback polling to finish
The VM takes up to 25.7 seconds to complete the nominal six-second retry loop. Allow 30 seconds for the three fallback waits and 40 seconds for the two shorter enclosing tests, retaining their lock-preservation and read-only assertions. The full daemon file is validated on Windows 11 and macOS.
* fix(daemon): retry transient Windows PID-file sharing violations
A transient EPERM replacing daemon.pid aborted startup, observed in the quiet-client test and reproduced with a real locked handle at
|
||
|
Colby Mchenry
|
1d62a67d8a |
docs(changelog): add entries for the Windows fixes (#2052)
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
1926ec39a2 |
fix(index): coordinate rebuilds with running MCP daemons (#1325) (#2051)
* fix: safely stop daemon before reindex * fix(index): coordinate rebuilds with running MCP daemons (#1325) Stop verified project daemons and confirm termination before recreating SQLite. Fence daemon and proxy reopening throughout rebuilds while preserving foreign and newer locks. Add Windows lifecycle coverage and cross-platform shutdown regressions. Co-authored-by: Harry B ✦ <drewavesx0@gmail.com> Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(mcp): fence project opens during index rebuilds after rebase Carry the #1325 rebuild fence through the upstream explicit-project lifecycle, read-only engine opens, synchronous initialization retries, and direct writer acquisition. Preserve shared project leases, read-only fallback behavior, freshness handling, and direct-mode query pools. Cover watched, unwatched, and read-only engines blocking during a foreign rebuild and recovering after the fence is released. * fix(mcp): answer a running index rebuild as guidance, not a tool error (#1325) A projectPath call during `codegraph index` hit assertNoRebuild's plain Error and returned isError; it is an expected, temporary condition, so it now uses a RebuildInProgressError that the tool handler answers success-shaped. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Harry B ✦ <drewavesx0@gmail.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
1b6be1c7d3 |
fix(watcher): ignore Windows access-only notifications (#1451) (#2050)
* fix Windows watcher false edits from NTFS atime * fix(watcher): ignore Windows access-only notifications (#1451) Check indexed size and mtime before recording Windows watcher edits. Preserve scope reconciliation and fail open when metadata cannot be verified. Add native Windows regression coverage and validate on Windows, macOS, and Linux. Co-authored-by: JJordan0K <69581081+JJordan0C@users.noreply.github.com> Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: JJordan0K <69581081+JJordan0C@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
346d3e3702 |
fix(mcp): terminate wedged local proxies with watchdog (#943) (#2049)
* fix(mcp): arm liveness watchdog in local proxy * fix(mcp): terminate wedged local proxies with watchdog (#943) Install the independent liveness watchdog in the local handshake proxy and stop it during shutdown. Build on PR #979 while preserving existing daemon assertions. Add process-termination, slow fallback, and watchdog cleanup coverage. Validate the repro and targeted tests on Windows, macOS, and Linux. Co-authored-by: Haoqian Li <haoqian.li@finalroundai.com> Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Haoqian Li <haoqian.li@finalroundai.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
d3671faf04 |
fix(installer): guide Windows shell users to PowerShell (#1294) (#2048)
Recognize MINGW, MSYS, and Cygwin and print the exact PowerShell installer command. Keep Darwin/Linux behavior unchanged and exit non-zero before downloading. Cover shell dispatch and real Windows ARM64 Git Bash behavior. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
1008b34e69 |
fix(bundle): ship a Windows Git Bash launcher (#1278) (#2047)
Ship an extensionless shell launcher alongside codegraph.cmd for Windows. Use bundled node.exe with matching runtime flags and preserve arguments, exit status, and inherited host PID. Validate on Windows 11, macOS, and Linux, including CLI/hooks and archive contents. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
7af240923d |
test(vitest): run pool workers with the WASM runtime flags (#1779) (#2046)
A vitest pool worker is the one launch path that never received
`--liftoff-only`, so the suite compiled tree-sitter grammars on the
turboshaft tier. Once ~640 extraction tests have warmed a grammar function
up, its background tier-up job (Turboshaft LoopUnrollingPhase, symbolized
from the native stack) exhausts a compiler Zone and aborts the worker:
`Fatal process out of memory: Zone`, which vitest reports only as "Worker
exited unexpectedly" with the file's remaining tests unrun.
Pass WASM_RUNTIME_FLAGS through poolOptions.forks.execArgv, so the test
process matches the bundled launcher, the CLI re-exec and refresh-launcher.
V8 flags are process-global, so parse worker threads are covered too.
Linux x64, Node 24.15, `__tests__/extraction.test.ts` (655 tests):
without the flag 3 of 4 runs died at the same test (639 passed);
with it 2 of 2 passed, in 19–24s instead of 34–41s. Neither
`--no-wasm-loop-unrolling` nor `--no-wasm-dynamic-tiering` prevents it.
(cherry picked from commit
|
||
|
Colby Mchenry
|
594bad7ec4 |
test(writer-lock): hold the lock with a live child, not a pid assumed alive (#2045)
The live-holder case wrote `pid: 1` and leaned on its own comment, "On Linux,
PID 1 is almost always alive". Windows has no PID 1, so tryAcquireWriterLock
read the holder as dead and acquired — the correct behaviour for a dead
holder, and a failing assertion. The platform-specific part was the fixture,
not the lock.
A parked child process is alive on every platform, and not stealing from a
live holder is what the lock actually promises, so the case now tests the
promise rather than a number that happens to name a running process. The child
is reaped in afterEach.
Fixes #1752.
(cherry picked from commit
|
||
|
Colby Mchenry
|
0a7321954c |
docs(changelog): add entries for the final wave of September 27 fixes (#2044)
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Colby Mchenry
|
2268a09b24 |
fix(sync): recover index freshness after lock contention (#1959) (#2043)
* fix(lock): keep a live writer lock regardless of file age (#1959) * fix(watcher): re-arm after lock contention and reconcile before trusting index (#1959) * feat(mcp): report exact index freshness without blocking the daemon (#1959) * fix(mcp): refuse drifted explore source after watcher degradation (#1959) * fix(mcp): refuse graph answers that name a drifted file after watcher degradation (#1959) Explore already refuses to render source from a file that changed while auto-sync was off. search, callers, callees and impact answer from the same frozen graph but kept replying, naming the changed file as if its edges were current. They now take the same refusal: the indexed files an answer names as locations are hashed against the index, and a drifted one is named instead of the answer. codegraph_node keeps its own drift gate, which serves the changed file's current bytes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(sync): recover index freshness after lock contention (#1959) Keep a live writer's lock regardless of its age. Re-arm lock-degraded watchers with throttling and retain stale state until full reconciliation succeeds. Validate contributing files through structured paths and hashes, refusing incomplete checks without tool errors. Report bounded freshness measurements and cover Git fallback exclusions. Co-authored-by: danusha2345 <danusha2345@users.noreply.github.com> Co-authored-by: danusha2345 <ewidusoc498@gmail.com> Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: danusha2345 <danusha2345@users.noreply.github.com> Co-authored-by: danusha2345 <ewidusoc498@gmail.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |