dirsql 0.4.20__tar.gz → 0.4.21__tar.gz

{dirsql-0.4.20 → dirsql-0.4.21}/Cargo.lock

@@ -499,7 +499,7 @@ dependencies = [
 
 [[package]]
 name = "dirsql-py-ext"
-version = "0.4.20"
+version = "0.4.21"
 dependencies = [
  "dirsql",
  "pyo3",

{dirsql-0.4.20 → dirsql-0.4.21}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dirsql
-Version: 0.4.20
+Version: 0.4.21
 Requires-Dist: tomli>=2 ; python_full_version < '3.11'
 Requires-Dist: bin-shim>=0.1
 Summary: Ephemeral SQL index over a local directory

{dirsql-0.4.20 → dirsql-0.4.21}/docs/.vitepress/config.ts

@@ -72,8 +72,7 @@ export default defineConfig({
             { text: 'Columns', link: '/reference/columns' },
             { text: 'Path-tables', link: '/reference/path-tables' },
             { text: 'HTTP API', link: '/reference/http-api' },
-            { text: 'SDK', link: '/reference/sdk' },
-            { text: 'Migrations', link: '/migrations' }
+            { text: 'SDK', link: '/reference/sdk' }
           ]
         },
         {

{dirsql-0.4.20/packages/python → dirsql-0.4.21}/docs/AGENTS.md

@@ -60,15 +60,13 @@ migrated -- a page survives only if a slot wants its content):
   protocol, exit codes, timeouts); the columns model (named tables get their
   hook's output; path-tables get stat columns); the
   HTTP API; the SDK page (`reference/sdk.md`, one page with
-  Python/TypeScript/Rust code-groups -- the sole SDK home); plus the
-  Migrations include (`migrations.md`).
+  Python/TypeScript/Rust code-groups -- the sole SDK home).
 - **Explanation** (`explanation.md`) -- one page: how `dirsql` thinks (the
   filesystem is the source of truth; the database is a derived, ephemeral,
   read-only view; reconcile and diffing). Its canonical home is the root
-  `ARCHITECTURE.md`;
-  #374 surfaces it via an include page, the same mechanism
-  `docs/migrations.md` uses for `MIGRATIONS.md`. Edit the root file, never
-  a rendered include.
+  `ARCHITECTURE.md`; #374 surfaces it via a VitePress include
+  (`<!--@include: ../ARCHITECTURE.md-->`). Edit the root file, never a
+  rendered include.
 
 Working rules:
 

{dirsql-0.4.20/packages/python → dirsql-0.4.21}/docs/howto/search-by-meaning.md

@@ -51,6 +51,7 @@ uvx --with dirsql-plugin-embeddings dirsql "
   SELECT path,
          vec_distance_cosine(emb, embed('how do I cook pasta?')) AS distance
   FROM (SELECT path, embed(content) AS emb FROM './notes/*.md')
+  WHERE emb IS NOT NULL
   ORDER BY distance
   LIMIT 3"
 ```
@@ -72,7 +73,12 @@ Reading the query inside-out:
    the one-liner normalizes a bare glob for you.
 2. `embed('how do I cook pasta?')` embeds the question once (the function is
    deterministic, so SQLite reuses the value across rows).
-3. `vec_distance_cosine(...)` computes cosine distance between the two
+3. `WHERE emb IS NOT NULL` drops the files that could not be embedded. A
+   file that is unreadable or not valid UTF-8 has
+   [`NULL` content](../reference/path-tables.md#columns), so its embedding
+   and its distance are NULL too — and SQLite sorts NULLs *first* ascending,
+   so without this line the unrankable files take the top-k slots.
+4. `vec_distance_cosine(...)` computes cosine distance between the two
    vectors; `ORDER BY distance LIMIT 3` keeps the three nearest.
 
 Structured files compose with SQL's JSON operators — embed one field instead
@@ -82,6 +88,7 @@ of the whole file:
 SELECT path
 FROM (SELECT path, embed(content ->> 'abstract') AS emb
       FROM './papers/**/metadata.json')
+WHERE emb IS NOT NULL
 ORDER BY vec_distance_cosine(emb, embed('local private models'))
 LIMIT 10
 ```

{dirsql-0.4.20/packages/rust → dirsql-0.4.21}/docs/plugins.md

@@ -35,6 +35,7 @@ uvx --with dirsql-plugin-embeddings dirsql "
   SELECT path
   FROM (SELECT path, embed(content ->> 'abstract') AS emb
         FROM './arxiv-firehose/data/**/metadata.json')
+  WHERE emb IS NOT NULL
   ORDER BY vec_distance_cosine(emb, embed('local private models'))
   LIMIT 10"
 ```

{dirsql-0.4.20 → dirsql-0.4.21}/packages/python/Cargo.toml

@@ -4,7 +4,7 @@ name = "dirsql-py-ext"
 # pypi/maturin handler can rewrite it via `write-version` before
 # `maturin build`. `pyproject.toml` declares `dynamic = ["version"]`
 # and maturin reads this field. Mirrors `packages/rust/Cargo.toml`.
-version = "0.4.20"
+version = "0.4.21"
 # abi3 note: the `abi3-py310` pyo3 feature (below) builds ONE stable-ABI
 # `cp310-abi3` wheel per platform that loads on every CPython >= 3.10
 # (matching `pyproject.toml`'s `requires-python = ">=3.10"`), instead of

{dirsql-0.4.20 → dirsql-0.4.21}/packages/python/MIGRATIONS.md

@@ -7,5 +7,4 @@ Migration entries for this package live as one file per breaking change under
 Nothing is assembled back into this file.
 
 Pre-fragment history (all packages) is frozen in the repository-root
-[`MIGRATIONS.md`](../../MIGRATIONS.md), surfaced on the docs site at
-[`/migrations`](https://thekevinscott.github.io/dirsql/migrations).
+[`MIGRATIONS.md`](../../MIGRATIONS.md).

{dirsql-0.4.20 → dirsql-0.4.21}/packages/python/docs/.vitepress/config.ts

@@ -72,8 +72,7 @@ export default defineConfig({
             { text: 'Columns', link: '/reference/columns' },
             { text: 'Path-tables', link: '/reference/path-tables' },
             { text: 'HTTP API', link: '/reference/http-api' },
-            { text: 'SDK', link: '/reference/sdk' },
-            { text: 'Migrations', link: '/migrations' }
+            { text: 'SDK', link: '/reference/sdk' }
           ]
         },
         {

{dirsql-0.4.20 → dirsql-0.4.21/packages/python}/docs/AGENTS.md

@@ -60,15 +60,13 @@ migrated -- a page survives only if a slot wants its content):
   protocol, exit codes, timeouts); the columns model (named tables get their
   hook's output; path-tables get stat columns); the
   HTTP API; the SDK page (`reference/sdk.md`, one page with
-  Python/TypeScript/Rust code-groups -- the sole SDK home); plus the
-  Migrations include (`migrations.md`).
+  Python/TypeScript/Rust code-groups -- the sole SDK home).
 - **Explanation** (`explanation.md`) -- one page: how `dirsql` thinks (the
   filesystem is the source of truth; the database is a derived, ephemeral,
   read-only view; reconcile and diffing). Its canonical home is the root
-  `ARCHITECTURE.md`;
-  #374 surfaces it via an include page, the same mechanism
-  `docs/migrations.md` uses for `MIGRATIONS.md`. Edit the root file, never
-  a rendered include.
+  `ARCHITECTURE.md`; #374 surfaces it via a VitePress include
+  (`<!--@include: ../ARCHITECTURE.md-->`). Edit the root file, never a
+  rendered include.
 
 Working rules:
 

{dirsql-0.4.20 → dirsql-0.4.21/packages/python}/docs/howto/search-by-meaning.md

@@ -51,6 +51,7 @@ uvx --with dirsql-plugin-embeddings dirsql "
   SELECT path,
          vec_distance_cosine(emb, embed('how do I cook pasta?')) AS distance
   FROM (SELECT path, embed(content) AS emb FROM './notes/*.md')
+  WHERE emb IS NOT NULL
   ORDER BY distance
   LIMIT 3"
 ```
@@ -72,7 +73,12 @@ Reading the query inside-out:
    the one-liner normalizes a bare glob for you.
 2. `embed('how do I cook pasta?')` embeds the question once (the function is
    deterministic, so SQLite reuses the value across rows).
-3. `vec_distance_cosine(...)` computes cosine distance between the two
+3. `WHERE emb IS NOT NULL` drops the files that could not be embedded. A
+   file that is unreadable or not valid UTF-8 has
+   [`NULL` content](../reference/path-tables.md#columns), so its embedding
+   and its distance are NULL too — and SQLite sorts NULLs *first* ascending,
+   so without this line the unrankable files take the top-k slots.
+4. `vec_distance_cosine(...)` computes cosine distance between the two
    vectors; `ORDER BY distance LIMIT 3` keeps the three nearest.
 
 Structured files compose with SQL's JSON operators — embed one field instead
@@ -82,6 +88,7 @@ of the whole file:
 SELECT path
 FROM (SELECT path, embed(content ->> 'abstract') AS emb
       FROM './papers/**/metadata.json')
+WHERE emb IS NOT NULL
 ORDER BY vec_distance_cosine(emb, embed('local private models'))
 LIMIT 10
 ```

{dirsql-0.4.20 → dirsql-0.4.21/packages/python}/docs/plugins.md

@@ -35,6 +35,7 @@ uvx --with dirsql-plugin-embeddings dirsql "
   SELECT path
   FROM (SELECT path, embed(content ->> 'abstract') AS emb
         FROM './arxiv-firehose/data/**/metadata.json')
+  WHERE emb IS NOT NULL
   ORDER BY vec_distance_cosine(emb, embed('local private models'))
   LIMIT 10"
 ```

{dirsql-0.4.20 → dirsql-0.4.21}/packages/rust/MIGRATIONS.md

@@ -7,5 +7,4 @@ Migration entries for this package live as one file per breaking change under
 Nothing is assembled back into this file.
 
 Pre-fragment history (all packages) is frozen in the repository-root
-[`MIGRATIONS.md`](../../MIGRATIONS.md), surfaced on the docs site at
-[`/migrations`](https://thekevinscott.github.io/dirsql/migrations).
+[`MIGRATIONS.md`](../../MIGRATIONS.md).

{dirsql-0.4.20 → dirsql-0.4.21}/packages/rust/docs/howto/search-by-meaning.md

@@ -51,6 +51,7 @@ uvx --with dirsql-plugin-embeddings dirsql "
   SELECT path,
          vec_distance_cosine(emb, embed('how do I cook pasta?')) AS distance
   FROM (SELECT path, embed(content) AS emb FROM './notes/*.md')
+  WHERE emb IS NOT NULL
   ORDER BY distance
   LIMIT 3"
 ```
@@ -72,7 +73,12 @@ Reading the query inside-out:
    the one-liner normalizes a bare glob for you.
 2. `embed('how do I cook pasta?')` embeds the question once (the function is
    deterministic, so SQLite reuses the value across rows).
-3. `vec_distance_cosine(...)` computes cosine distance between the two
+3. `WHERE emb IS NOT NULL` drops the files that could not be embedded. A
+   file that is unreadable or not valid UTF-8 has
+   [`NULL` content](../reference/path-tables.md#columns), so its embedding
+   and its distance are NULL too — and SQLite sorts NULLs *first* ascending,
+   so without this line the unrankable files take the top-k slots.
+4. `vec_distance_cosine(...)` computes cosine distance between the two
    vectors; `ORDER BY distance LIMIT 3` keeps the three nearest.
 
 Structured files compose with SQL's JSON operators — embed one field instead
@@ -82,6 +88,7 @@ of the whole file:
 SELECT path
 FROM (SELECT path, embed(content ->> 'abstract') AS emb
       FROM './papers/**/metadata.json')
+WHERE emb IS NOT NULL
 ORDER BY vec_distance_cosine(emb, embed('local private models'))
 LIMIT 10
 ```

{dirsql-0.4.20/packages/python → dirsql-0.4.21/packages/rust}/docs/plugins.md

@@ -35,6 +35,7 @@ uvx --with dirsql-plugin-embeddings dirsql "
   SELECT path
   FROM (SELECT path, embed(content ->> 'abstract') AS emb
         FROM './arxiv-firehose/data/**/metadata.json')
+  WHERE emb IS NOT NULL
   ORDER BY vec_distance_cosine(emb, embed('local private models'))
   LIMIT 10"
 ```

dirsql-0.4.20/docs/migrations.md

@@ -1,5 +0,0 @@
----
-title: Migrations
----
-
-<!--@include: ../MIGRATIONS.md-->

dirsql-0.4.20/packages/python/docs/migrations.md

@@ -1,5 +0,0 @@
----
-title: Migrations
----
-
-<!--@include: ../MIGRATIONS.md-->

dirsql-0.4.20/packages/rust/docs/migrations.md

@@ -1,5 +0,0 @@
----
-title: Migrations
----
-
-<!--@include: ../MIGRATIONS.md-->