SNOWFLAKE DOCS DIFF
本日は全8,375ページ中22件の改訂があり、Declarative Native App のノートブック共有基盤が Legacy Notebooks から Workspaces へ全面移行されたことが最大の変更。加えて、データ分類プロファイルへの AI mode 追加（GPT-5 Mini 連携）と IS_ROLE_ACTIVATED 関数の複数ロール対応というセキュリティ・ガバナンス領域の機能拡張が目立つ。破壊的変更はないものの、Legacy Notebooks の新規作成停止が本日付で発効しており、該当プロバイダーは対応が必要。
ai_mode パラメータが追加され、LLM による高精度なセマンティック分類が可能になったが、AI_SENSITIVE_DATA_CLASSIFICATION として追加課金が発生する。Declarative Native App でのノートブック共有が、従来の Legacy Notebooks（application_content.notebooks）から Snowflake Notebooks in Workspaces ベースに全面移行された。ページタイトル・説明文・ナビゲーションリンクが刷新され、「Notebook Live Editing」「Development Mode」といった旧用語が削除され、ワークスペース内での直接編集・公開フローの説明に置き換えられた。
影響: Legacy Notebooks 方式での新規パッケージ作成は 2026年8月18日以降不可となるため、既存プロバイダーは Workspaces への移行が必要。
判定根拠: SQL 構文/コードブロックの増減 (+2/-7)
--- ahttps://docs.snowflake.com/en/developer-guide/declarative-sharing/live-editing+++ bhttps://docs.snowflake.com/en/developer-guide/declarative-sharing/live-editing@@ -1,98 +1,99 @@-# Editing Notebooks in Declarative Shared Native Applications+# Editing notebooks in a %declarative-app-short% -Sharing individual notebooks with `application_content.notebooks` shares-[Legacy Notebooks](/user-guide/ui-snowsight/notebooks), and is deprecated. Starting-August 18, 2026, you can't create new application packages that share notebooks this way. Starting-November 2026, Legacy Notebooks can no longer be run or edited. For the full timeline, see-[](/release-notes/bcr-bundles/un-bundled/bcr-disable-legacy-notebooks).--Share [Snowflake Notebooks in Workspaces](/user-guide/ui-snowsight/notebooks-in-workspaces/notebooks-in-workspaces-overview)-in a workspace instead. Consumers can run and interact with those notebooks the same way, and a-workspace can also share documentation, images, and sample data files. For more information, see-[](/developer-guide/declarative-sharing/workspaces).+Workspace sharing in %declarative-app-short-plural% is available to all accounts. - [](/developer-guide/declarative-sharing/about) - [](/developer-guide/declarative-sharing/introduction)+- [](/developer-guide/declarative-sharing/workspaces) - [](/developer-guide/declarative-sharing/package)-- [](/developer-guide/declarative-sharing/udfs-sprocs)-- [](/developer-guide/declarative-sharing/monitoring) - [](/developer-guide/declarative-sharing/versioning)-- [](/developer-guide/declarative-sharing/listing) - [](/developer-guide/declarative-sharing/tutorials/getting-started)-- [](/developer-guide/declarative-sharing/command-reference)-- [](/developer-guide/declarative-sharing/app-roles) - [](/developer-guide/declarative-sharing/manifest-reference)-- [](/developer-guide/declarative-sharing/add-packages)-- [](/developer-guide/declarative-sharing/consumer/install)-- [](/developer-guide/declarative-sharing/consumer/access-app-content)-- [](/developer-guide/declarative-sharing/consumer/consumer-execution)+- [](/developer-guide/declarative-sharing/consumer/access-shared-workspace) - [](/developer-guide/declarative-sharing/limitations) Support for %declarative-app-plural% is available to all accounts. -%declarative-app-short-plural% can include [notebooks](/user-guide/ui-snowsight/notebooks) to query, visualize, and explore the data. This topic describes how to use **Notebook Live Editing** to streamline the development and testing of notebooks within an application.+%declarative-app-short-plural% share [Snowflake Notebooks in Workspaces](/user-guide/ui-snowsight/notebooks-in-workspaces/notebooks-in-workspaces-overview)+in a [workspace](/developer-guide/declarative-sharing/workspaces). As a provider, you can edit those+notebooks in place while you develop the app, instead of editing them elsewhere and rebuilding the+application package for every change. -Using the **Notebook Live Editing** feature, you can speed up the development process by editing and testing notebooks directly within an application. This saves you from having to develop notebooks externally, or from having to rebuild the application package for every change.+To do this, you create a test application from the live version of your application package. Notebooks+in a workspace in that test application are editable, so you can change a notebook, run it to check+the result, and publish the change when you're satisfied. Publishing writes the change back to the+application package, so the next version you release includes it. -**Notebook Live Editing** uses a **Development Mode** that lets you make changes to notebooks "live" inside an application instance. Your edits are saved to a dedicated live version of the application package, allowing for rapid, on-the-fly-testing and iteration.+Consumers who install your app always get a read-only workspace. Editing applies only to a test+application you create from your own package. -<a id="label-dsna-how-it-works"></a>+<a id="label-dsna-live-editing-how-it-works"></a> -## How It Works+## How it works -The workflow uses a live version of your application package, which functions as a development sandbox. This tutorial describes how to set up-and use the Notebook Live Editing feature.+The live version of your application package acts as a development sandbox. You edit notebooks in a+test application, publish the changes you want to keep, and then release the live version when the+app is ready. -<a id="label-dsna-step-1-set-up-the-development-environment"></a>+<a id="label-dsna-live-editing-create-app"></a> -### Step 1: Set Up the Development Environment+### Step 1: Create a test application from the live version -To begin, you need a package that contains the following:--- A manifest file that defines the application and its components.-- A notebook that you can edit and test.--You then create an application instance from the live version of your package. The live version is created automatically when you create your application.--1. Build the package.- ```sql- ALTER APPLICATION PACKAGE pkg_name BUILD;- ```--2. Create an application instance from the live version. Notebooks in this new application will automatically be in **Development- Mode**, allowing live editing. Prior to this step, notebooks in the application are in **Read-only mode**.- ```sql- CREATE APPLICATION live_app_name- FROM APPLICATION PACKAGE pkg_name- USING VERSION LIVE;- ```--<a id="label-dsna-step-2-live-edit-and-test-notebooks"></a>--### Step 2: Live Edit and Test Notebooks--With your `live_app_name` application running, in SnowSight, open your app from the list of apps in your account, and open one of its notebooks from its listing page. After creating the application from the application package in the previous step, the applications' notebooks will be in **Developer mode**.You can now do the following:--- Edit notebook cells directly in the browser.-- Run and test your code immediately within the context of the application.--Any changes you make are instantly saved to the live version of the `pkg_name` application package. This allows you to iterate changes-to your application quickly, without needing to perform a full package build for each minor adjustment.--<a id="label-dsna-step-3-finalize-and-release-changes"></a>--### Step 3: Finalize and Release Changes--Once you're happy with the state of your notebooks, you can promote the live version to a stable release. This freezes the current state of-your notebooks and makes them part of a permanent application version. The app framework automatically creates a version number for your-release.--- Release the live version to finalize your work.+Build the package, and then create an application from its live version: ```sql-ALTER APPLICATION PACKAGE pkg_name RELEASE LIVE VERSION;+ALTER APPLICATION PACKAGE <pkg_name> BUILD;++CREATE APPLICATION <test_app_name>+ FROM APPLICATION PACKAGE <pkg_name>+ USING VERSION LIVE; ``` -This command creates a new, immutable version of your application package containing all the notebook changes you made. For more information-about the application package and the live version, see [](/developer-guide/declarative-sharing/package).+Notebooks in the workspaces of this application are editable. For more information about building and+releasing a package, see [](/developer-guide/declarative-sharing/package).++<a id="label-dsna-live-editing-edit"></a>++### Step 2: Edit a notebook in the workspace++1. Sign in to %sf-web-interface-link%.+2. In the navigation menu, select **Apps**.+3. Select your test application. The workspaces that the application contains are listed.+4. Select the workspace, and then open an `.ipynb` file.+5. Edit the notebook cells and run them to test your changes. ... (truncated)
差分が長いため、途中まで表示しています。
Declarative Native Apps チュートリアルで、共有コンテンツの形式が個別のノートブックファイルからワークスペースディレクトリ（app/workspace）に全面的に置き換えられた。冒頭にあった Legacy Notebooks の非推奨警告と移行タイムライン（2026年8月18日以降の新規作成停止、11月のサポート終了）が削除され、「Workspace sharing is available to all accounts」という簡潔な記述に変更された。また、目次に /developer-guide/declarative-sharing/workspaces へのリンクが追加され、manifest.yml に application_content.workspaces セクションの YAML 例が新設された。
影響: 今後このチュートリアルに沿って Declarative Native App を作成する場合、ノートブック単体ではなくワークスペースディレクトリ構成で作業する必要があり、既存手順をそのまま流用している場合はmanifest.yml と配置構造の見直しが必要になる。
判定根拠: SQL 構文/コードブロックの増減 (+5/-1)
--- ahttps://docs.snowflake.com/en/developer-guide/declarative-sharing/tutorials/getting-started+++ bhttps://docs.snowflake.com/en/developer-guide/declarative-sharing/tutorials/getting-started@@ -1,18 +1,10 @@ # Tutorial: Getting started with %declarative-app-short-plural% -Sharing individual notebooks with `application_content.notebooks` shares-[Legacy Notebooks](/user-guide/ui-snowsight/notebooks), and is deprecated. Starting-August 18, 2026, you can't create new application packages that share notebooks this way. Starting-November 2026, Legacy Notebooks can no longer be run or edited. For the full timeline, see-[](/release-notes/bcr-bundles/un-bundled/bcr-disable-legacy-notebooks).--Share [Snowflake Notebooks in Workspaces](/user-guide/ui-snowsight/notebooks-in-workspaces/notebooks-in-workspaces-overview)-in a workspace instead. Consumers can run and interact with those notebooks the same way, and a-workspace can also share documentation, images, and sample data files. For more information, see-[](/developer-guide/declarative-sharing/workspaces).+Workspace sharing in %declarative-app-short-plural% is available to all accounts. - [](/developer-guide/declarative-sharing/about) - [](/developer-guide/declarative-sharing/introduction) - [](/developer-guide/declarative-sharing/package)+- [](/developer-guide/declarative-sharing/workspaces) - [](/developer-guide/declarative-sharing/live-editing) - [](/developer-guide/declarative-sharing/monitoring)@@ -32,5 +24,5 @@ This tutorial takes Snowflake data providers through the process of creating, publishing, and accessing a %declarative-app%. The tutorial uses [SnowCLI](/developer-guide/snowflake-cli/index) as well-as a provided notebook file and a partially complete manifest file to create a %declarative-app-short%.+as a provided workspace directory and a partially complete manifest file to create a %declarative-app-short%. This tutorial includes two personas:@@ -114,5 +106,5 @@ - SQL files for creating all required artifacts. These files can be used to speed the process of setting up and tearing down your tutorial environment.-- A notebook file that contains the logic for the %declarative-app-short%.+- A workspace directory (*app/workspace*) that contains the content shared by the %declarative-app-short%: a notebook (`.ipynb`) file and a readme. - A manifest file that contains the metadata for the %declarative-app-short%, which you will need to make minor modifications to during the tutorial. - A sample configuration file for SnowCLI, which you can use to configure your Snowflake connection.@@ -287,6 +279,7 @@ As a provider, create and package the %declarative-app-short%. Note that this step uses: -- A provided NOTEBOOK file that contains the logic for the %declarative-app-short%.- This notebook contains a single SQL statement that queries the table created in the previous step.+- A provided workspace directory, *app/workspace*, that contains the content shared by the %declarative-app-short%.+ The directory holds a notebook that queries the table created in the previous step, and a readme that describes+ what the app shares. Creating a %declarative-app-short% involves:@@ -295,5 +288,5 @@ A starting point for the manifest is provided in the `manifest.yml` file. 2. Creating the %declarative-app-short% package.-3. Packaging the app with its manifest and the associated notebook.+3. Packaging the app with its manifest and the associated workspace directory. 4. Validating the %declarative-app-short% package. @@ -342,5 +335,20 @@ ### Populate the application package -Next, populate the app package with the manifest and notebook files.+Next, populate the app package with the manifest file and the workspace files.++The `manifest.yml` file provided with this tutorial declares a workspace whose `source` is the+*workspace/* directory:++```yaml+application_content:+ workspaces:+ - DS_WORKSPACE:+ source: workspace/+ comment: My first DS app workspace.+```++Because `source` is `workspace/`, the workspace files must be uploaded into a *workspace/* directory+of the package version. Uploading them to the root of the version makes the build fail, because the+declared source directory would be empty. **%sf-web-interface%:**@@ -350,6 +358,7 @@ 1. In %sf-web-interface%, navigate to the listing for the app package you created in the previous step. 2. Select **Manage files**, and then select **Upload files**.-3. Drag the notebook files and the manifest from the *tmp/tutorial/app* folder to the **Upload files** dialog where indicated, or select **Browse** to locate and select the files.-4. Select **Upload** to upload the files to the live stage and trigger a build.+3. Drag the manifest from the *tmp/tutorial/app* folder to the **Upload files** dialog where indicated, or select **Browse** to locate and select the file.+4. Drag the *workspace* folder from *tmp/tutorial/app* to the dialog. The folder structure is preserved, so the files land in a *workspace/* directory.+5. Select **Upload** to upload the files to the live stage and trigger a build. If the build is successful, the **Live version** tab on the app package's listing displays the following:@@ -357,5 +366,5 @@ - The last build time - The contents of the last build, including:- - A list of the app package's notebooks+ - A list of the app package's workspaces - A list of shared objects @@ -366,15 +375,18 @@ To use SQL in Snowflake CLI to populate the app package: -1. First, add a notebook using the command below. For more information see [](/sql-reference/sql/put).-- The notebook file provided with this tutorial contains a single SQL statement that queries the previously created table.--- This command assumes the notebook is named `NOTEBOOK.ipynb` and is located in the `/tmp/tutorial/app` folder. You may need to modify the path to the notebook file based on where you unzipped the tutorial files.--- ```snowcli- snow sql -q "PUT file:////tmp/tutorial/app/NOTEBOOK.ipynb \- snow://package/<DECL_SHARE_APP_PKG>/versions/LIVE/ \+1. First, add the workspace files using the commands below. For more information see [](/sql-reference/sql/put).++ The notebook provided with this tutorial queries the table you created previously. The readme+ describes what the app shares.+++ These commands assume the workspace files are in the `/tmp/tutorial/app/workspace` folder. You may need to modify the paths based on where you unzipped the tutorial files.++ Note the `workspace/` suffix on the target URL. It must match the `source` value in the manifest file.+++ ```snowcli+ snow sql -q "PUT file:////tmp/tutorial/app/workspace/NOTEBOOK.ipynb \+ snow://package/<DECL_SHARE_APP_PKG>/versions/LIVE/workspace/ \ OVERWRITE=TRUE AUTO_COMPRESS=false;" ```@@ -386,6 +398,14 @@ | source | target | source_size | target_size | source_compression | target_compression | status | message | |----------------+----------------+-------------+-------------+--------------------+--------------------+----------+---------|- | NOTEBOOK.ipynb | NOTEBOOK.ipynb | 817 | 832 | NONE | NONE | UPLOADED | |+ | NOTEBOOK.ipynb | NOTEBOOK.ipynb | 1777 | 1792 | NONE | NONE | UPLOADED | | +----------------------------------------------------------------------------------------------------------------------------++ ```++ Then add the readme:++ ```snowcli+ snow sql -q "PUT file:////tmp/tutorial/app/workspace/README.md \+ snow://package/<DECL_SHARE_APP_PKG>/versions/LIVE/workspace/ \+ OVERWRITE=TRUE AUTO_COMPRESS=false;" ``` @@ -435,5 +455,5 @@ ### Version and release the app -Once an app package has been created and populated with a notebook and manifes... (truncated)
差分が長いため、途中まで表示しています。
分類プロファイルで AI mode を有効化する新セクションが追加された。プロファイル作成時に 'ai_mode': true を設定するか、既存プロファイルに対して SET_AI_MODE メソッドを呼び出すことで LLM を使った追加のセマンティックカテゴリ識別が可能になる。現時点でサポートされるモデルは OpenAI GPT-5 Mini のみ。Enterprise Edition 以上が必要。
影響: データ分類の精度向上を目的に AI mode を導入する場合、プロファイルの新規作成または既存プロファイルへの SET_AI_MODE 呼び出しが必要となり、Enterprise Edition 以上のライセンスが前提となる。
判定根拠: SQL 構文/コードブロックの増減 (+9/-0)
--- ahttps://docs.snowflake.com/en/user-guide/classify-auto+++ bhttps://docs.snowflake.com/en/user-guide/classify-auto@@ -41,4 +41,37 @@ For an example of using the CREATE CLASSIFICATION_PROFILE command to create a classification profile, see [](#label-classify-auto-example).++<a id="label-classify-auto-ai-mode"></a>++## Use AI mode in a classification profile++Available to Enterprise Edition or higher.++You can enable [AI mode](/user-guide/classify-intro#label-classify-ai-mode) on a classification profile to use LLMs to identify additional+semantic categories beyond those identified by standard classification. Set `'ai_mode': true` in the profile configuration when you create+the profile:++```sql+CREATE OR REPLACE SNOWFLAKE.DATA_PRIVACY.CLASSIFICATION_PROFILE+ my_ai_profile({+ 'minimum_object_age_for_classification_days': 0,+ 'auto_tag': true,+ 'ai_mode': true+ });+```++Currently, AI mode only supports the OpenAI GPT-5 Mini model.++To enable AI mode on an existing classification profile, call the `SET_AI_MODE` method:++```sql+CALL my_ai_profile!SET_AI_MODE(true);+```++To disable AI mode on an existing profile:++```sql+CALL my_ai_profile!SET_AI_MODE(false);+``` <a id="label-classify-auto-map-tags"></a>
IS_ROLE_ACTIVATED 関数（SYS_CONTEXT 経由）が、1回の呼び出しで複数ロールをまとめてチェックできるよう拡張された。構文が '<role>' [ , '<role>' ... ] に変更され、指定したロールのいずれかがアクティブであれば 'TRUE' を返す。ただし複数ロール指定は定数値のみで可能であり、行アクセスポリシー内のカラム参照などの式では単一ロールのみ指定可能。
影響: 複数ロールの OR 判定を1回の関数呼び出しで行えるようになり、セキュリティポリシーや条件分岐の記述が簡潔になる。
判定根拠: SQL 構文/コードブロックの増減 (+3/-0)
--- ahttps://docs.snowflake.com/en/sql-reference/functions/is_role_activated+++ bhttps://docs.snowflake.com/en/sql-reference/functions/is_role_activated@@ -10,6 +10,6 @@ # IS_ROLE_ACTIVATED (SYS_CONTEXT function) -Returns the VARCHAR value `'TRUE'` if an account role is activated. You can check activation in the session context or in the-current execution context.+Returns the VARCHAR value `'TRUE'` if an account role is activated. You can check one role or several roles in a single call,+and you can check activation in the session context or in the current execution context. <dl>@@ -32,5 +32,5 @@ 'SNOWFLAKE$SESSION' , 'IS_ROLE_ACTIVATED' ,- '<role>'+ '<role>' [ , '<role>' ... ] ) ```@@ -42,5 +42,5 @@ 'SNOWFLAKE$CURRENT' , 'IS_ROLE_ACTIVATED' ,- '<role>'+ '<role>' [ , '<role>' ... ] ) ```@@ -69,8 +69,9 @@ </dd>-<dt><code className="samp">'<em>role</em>'</code></dt>+<dt><code className="samp">'<em>role</em>' [ , '<em>role</em>' ... ]</code></dt> <dd> -Specifies the account role to check.+Specifies the account role or roles to check. You can pass one role or several roles as separate arguments. The function returns+`'TRUE'` if any of the specified roles is activated. Multiple roles must be constant values (see the usage notes). </dd>@@ -81,6 +82,6 @@ The function returns one of the following VARCHAR values: -- `'TRUE'` if the account role is activated in the specified context.-- `'FALSE'` if the account role is not activated or if the account role is not valid.+- `'TRUE'` if any of the specified account roles is activated in the specified context.+- `'FALSE'` if none of the specified account roles is activated, or if none of the specified roles exists. To compare this return value against the BOOLEAN value TRUE or FALSE, [cast](#label-data-type-explicit-casting) the return@@ -108,4 +109,10 @@ stored procedure, for example, this reflects the owner's activated roles, not the caller's. +- When you specify more than one role, the function returns `'TRUE'` if any of the roles is activated. Snowflake ignores+ duplicate role names.++- You can specify multiple roles only when each role is a constant value, such as a string literal. When the role is an+ expression that isn't a constant, such as a column reference in a row access policy, you can specify only a single role.+ - To simulate the result of this function in a policy, use the `SNOWFLAKE$SESSION_ACTIVATED_ROLES` or `SNOWFLAKE$CURRENT_ACTIVATED_ROLES` [list argument](/sql-reference/functions/policy_context#label-policy-context-sys-context-list-args)@@ -126,2 +133,9 @@ SELECT SYS_CONTEXT('SNOWFLAKE$CURRENT', 'IS_ROLE_ACTIVATED', 'analyst'); ```++The following example checks several roles in a single call. It returns `'TRUE'` if any of the roles `analyst`, `auditor`, or+`reviewer` is activated in the session:++```sql+SELECT SYS_CONTEXT('SNOWFLAKE$SESSION', 'IS_ROLE_ACTIVATED', 'analyst', 'auditor', 'reviewer');+```
CREATE CLASSIFICATION_PROFILE に新パラメータ ai_mode（BOOLEAN、デフォルト FALSE）が追加された。TRUE に設定すると LLM を使用して標準分類では検出できない追加のセマンティックカテゴリを識別する「AI mode」が有効になる。LLM の利用は AI_SENSITIVE_DATA_CLASSIFICATION サービスタイプで課金される。関連メソッド SET_AI_MODE と AI mode 有効化の SQL 例も追加された。
影響: AI mode を有効化すると分類精度が向上する一方、AI_SENSITIVE_DATA_CLASSIFICATION として追加課金が発生するため、コストへの影響を事前に確認する必要がある。
判定根拠: SQL 構文/コードブロックの増減 (+4/-0)
--- ahttps://docs.snowflake.com/en/sql-reference/classes/classification_profile/commands/create-classification-profile+++ bhttps://docs.snowflake.com/en/sql-reference/classes/classification_profile/commands/create-classification-profile@@ -123,4 +123,20 @@ </tr> <tr>+ <td>`ai_mode`</td>+ <td>BOOLEAN</td>+ <td>FALSE</td>+ <td>++Optional: When `TRUE`, enables [AI mode](/user-guide/classify-intro#label-classify-ai-mode) to use LLMs to identify additional semantic+categories beyond those identified by standard classification.++When `FALSE`, only standard classification is used.++LLM usage is billed under the `AI_SENSITIVE_DATA_CLASSIFICATION` [service type](/sql-reference/service-types). Serverless compute for+classification continues to be billed under `SENSITIVE_DATA_CLASSIFICATION`.++</td>+ </tr>+ <tr> <td>`auto_tag`</td> <td>BOOLEAN</td>@@ -273,4 +289,5 @@ - [<classification_profile_name>!DESCRIBE](/sql-reference/classes/classification_profile/methods/describe)+- [<classification_profile_name>!SET_AI_MODE](/sql-reference/classes/classification_profile/methods/set_ai_mode) - [<classification_profile_name>!SET_AUTO_TAG](/sql-reference/classes/classification_profile/methods/set_auto_tag) - [<classification_profile_name>!SET_CLASSIFY_VIEWS](/sql-reference/classes/classification_profile/methods/set_classify_views)@@ -372,2 +389,14 @@ ); ```++Create an instance with [AI mode](/user-guide/classify-intro#label-classify-ai-mode) enabled:++```sql+CREATE OR REPLACE SNOWFLAKE.DATA_PRIVACY.CLASSIFICATION_PROFILE my_ai_profile(+ {+ 'minimum_object_age_for_classification_days': 0,+ 'auto_tag': true,+ 'ai_mode': true+ }+);+```
2026年7月〜8月のパフォーマンス改善項目が多数追加された。主な内容は、Dynamic Table関連の強化（ADAPTIVE refresh mode の GA、Custom incremental の GA、複数テーブル一括 REFRESH、VOLATILE UDF 対応）、Optima Planning の改善（プランシェイプによるクエリマッチング、動的サンプリングによるフィルタ選択性推定）、およびコンパイル・実行の最適化（CASE式の共通部分式の重複計算排除、大規模 disjunctive tuple filter の CTE semi-join 書き換え、IFF/CASE式のプルーニング改善、ワーカープロセス再利用、CTE配下へのフィルタプッシュダウン）。 影響: Dynamic Table のリフレッシュ戦略選択が自動化（ADAPTIVE が推奨デフォルト）され、Optima Planning の適用範囲が拡大したことで、BI ダッシュボードや定型ワークロードのチューニング負荷が軽減される。
変更内容: 本文を更新（144行追加・1行削除）
--- ahttps://docs.snowflake.com/en/release-notes/performance-improvements-2026+++ bhttps://docs.snowflake.com/en/release-notes/performance-improvements-2026@@ -22,8 +22,151 @@ <tbody> <tr>+ <td>August 2026</td>+ <td>++[Optima Planning](/user-guide/snowflake-optima) now matches queries by plan shape instead of exact+query text, so execution feedback applies across structurally similar queries, not just identical+repeats.++</td>+ <td>++Speeds up recurring BI dashboard and application-generated SQL that varies query to query but shares+the same join pattern, broadening which workloads benefit from Optima Planning.++</td>+ </tr>+ <tr>+ <td>July 2026</td>+ <td>++[Dynamic table](/user-guide/dynamic-tables/overview)+[ADAPTIVE refresh mode](/user-guide/dynamic-tables/refresh-modes#label-dynamic-tables-refresh-adaptive)+(*General availability*). Automatically reinitializes a dynamic table with a full refresh when+upstream change volume makes incremental refresh significantly more expensive, then resumes+incremental refresh.++</td>+ <td>++Removes the need to manually choose between incremental and full refresh. Recommended as the default+refresh mode for incremental workloads.++</td>+ </tr>+ <tr>+ <td>July 2026</td>+ <td>++[Custom incremental dynamic tables](/user-guide/dynamic-tables/custom-incrementalization)+(*General availability*).++</td>+ <td>++Lets more dynamic table pipelines stay on efficient incremental refresh instead of falling back to+full refresh for query patterns that weren't previously supported.++</td>+ </tr>+ <tr>+ <td>July 2026</td>+ <td>++Refresh multiple [dynamic tables](/user-guide/dynamic-tables/overview) in a single+`ALTER DYNAMIC TABLE ... REFRESH` statement. Shared upstreams refresh exactly once.++</td>+ <td>++Reduces redundant refresh compute when manually refreshing several related dynamic tables together.++</td>+ </tr>+ <tr>+ <td>July 2026</td>+ <td>++VOLATILE scalar user-defined functions (UDFs) are supported with [dynamic table](/user-guide/dynamic-tables/overview)+incremental refresh in INCREMENTAL, AUTO, and ADAPTIVE refresh modes (*General availability*).+Applies to UDFs written in Python, Java, Scala, JavaScript, and SQL.++</td>+ <td>++More dynamic table pipelines that use VOLATILE UDFs can stay on incremental refresh instead of full+refresh.++</td>+ </tr>+ <tr>+ <td>July 2026</td>+ <td>++[Optima Planning](/user-guide/snowflake-optima) uses dynamic sampling to sharpen filter-selectivity+estimates for recurring workloads.++</td>+ <td>++More accurate cardinality estimates produce better-informed query plans for recurring workloads.++</td>+ </tr>+ <tr>+ <td>July 2026</td>+ <td>Shared common sub-expressions in CASE expressions are computed once instead of repeatedly.</td>+ <td>Reduces compilation time for queries with large or repeated CASE expressions.</td>+ </tr>+ <tr>+ <td>July 2026</td>+ <td>Large disjunctive tuple filters are rewritten as CTE semi-joins during compilation.</td>+ <td>Faster compilation for queries with large disjunctive tuple filters.</td>+ </tr>+ <tr>+ <td>July 2026</td>+ <td>Improved pruning for queries that rely on IFF and CASE expressions.</td>+ <td>Faster compilation for queries that rely on generic partition pruning.</td>+ </tr>+ <tr>+ <td>July 2026</td>+ <td>Query worker processes are reused across queries instead of starting fresh for each query.</td>+ <td>Lower latency for short-running queries.</td>+ </tr>+ <tr>+ <td>July 2026</td>+ <td>Pushes filters below CTEs so less intermediate data is materialized.</td>+ <td>Faster execution for queries that filter data produced by a CTE.</td>+ </tr>+ <tr>+ <td>June 2026</td>+ <td>++[Adaptive Compute](/user-guide/warehouses-adaptive) (*General availability*). Create Adaptive+Warehouses or convert existing standard warehouses to Adaptive Warehouses without downtime.++</td>+ <td>++Automatically tunes warehouse compute to each query's needs, improving price-performance for variable+workloads without manual sizing.++</td>+ </tr>+ <tr>+ <td>June 2026</td>+ <td>The plan cost model accounts for join-filter cost, producing better join orders.</td>+ <td>Faster execution for join-heavy recurring workloads through improved join ordering.</td>+ </tr>+ <tr>+ <td>June 2026</td>+ <td>Improved automatic rewriting for materialized views.</td>+ <td>Faster compilation for materialized-view-rewrite-eligible workloads, with no missed rewrites.</td>+ </tr>+ <tr> <td>May 2026</td> <td> -[Snowflake Optima Planning](/user-guide/snowflake-optima) (*General availability*). Automatically+[Optima Planning](/user-guide/snowflake-optima) (*General availability*). Automatically improves query plans for recurring workloads by learning from past executions. No configuration required.
background パラメータの説明が改訂され、stream パラメータとの組み合わせによる動作が明確化された。background=true かつ stream=false の場合は即座に run_id を返し、stream=true の場合は Server-Sent Events でストリーミングされる旨が追記された。また warehouse パラメータに未指定時はユーザーのデフォルトウェアハウスが使用されるという説明が追加された。
影響: background と stream の組み合わせパターンが明文化されたことで、非同期実行時のクライアント実装判断がしやすくなり、warehouse 未指定時の挙動も把握できるようになる。
変更内容: 本文を更新（21行追加・20行削除）
--- ahttps://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-run+++ bhttps://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-run@@ -141,13 +141,13 @@ <td> Whether to run the agent asynchronously. If `true`, the agent runs- asynchronously and returns immediately with `"status": "in_progress"`- and a `run_id`. The run continues in the background with a 6 hour- timeout, even if the client disconnects. Retrieve the response after- the run completes by streaming from the Stream Agent Run endpoint with- the `run_id` (REST API) or by polling- [](/sql-reference/functions/thread_messages-snowflake-cortex) with the- thread ID (SQL function). If `false`, the agent runs synchronously with- a 15 minute timeout. Only available when using threads to manage- conversation history.+ asynchronously in the background with a 6 hour timeout, even if the+ client disconnects. For a background run with `stream` set to `false`,+ the API returns immediately with `status: in_progress` and a `run_id`;+ retrieve the response after the run completes by streaming from the+ Stream Agent Run endpoint with the `run_id` (REST API) or by polling the+ THREAD_MESSAGES SQL function with the thread ID. If `stream` is `true`,+ the response is streamed as Server-Sent Events. If `background` is+ `false`, the agent runs synchronously with a 15 minute timeout. Only+ available when using threads to manage conversation history. </td> </tr>@@ -316,13 +316,13 @@ <td> Whether to run the agent asynchronously. If `true`, the agent runs- asynchronously and returns immediately with `"status": "in_progress"`- and a `run_id`. The run continues in the background with a 6 hour- timeout, even if the client disconnects. Retrieve the response after- the run completes by streaming from the Stream Agent Run endpoint with- the `run_id` (REST API) or by polling- [](/sql-reference/functions/thread_messages-snowflake-cortex) with the- thread ID (SQL function). If `false`, the agent runs synchronously with- a 15 minute timeout. Only available when using threads to manage- conversation history.+ asynchronously in the background with a 6 hour timeout, even if the+ client disconnects. For a background run with `stream` set to `false`,+ the API returns immediately with `status: in_progress` and a `run_id`;+ retrieve the response after the run completes by streaming from the+ Stream Agent Run endpoint with the `run_id` (REST API) or by polling the+ THREAD_MESSAGES SQL function with the thread ID. If `stream` is `true`,+ the response is streamed as Server-Sent Events. If `background` is+ `false`, the agent runs synchronously with a 15 minute timeout. Only+ available when using threads to manage conversation history. </td> </tr>@@ -2102,6 +2102,7 @@ <td>string</td> <td>- The name of the warehouse. Case-sensitive, if it is an unquoted- identifier, provide the name in all-caps.+ The name of the warehouse. Case-sensitive; if it is an unquoted+ identifier, provide the name in all-caps. If not specified, the+ user's default warehouse is used. </td> </tr>
機密データ分類の分類プロファイルに AI mode が新規追加された。AI mode を有効にすると openai-gpt-5-mini モデル（LLM）を使用して、標準分類では検出できない追加のセマンティックカテゴリを識別し精度を向上させる。有効化は Trust Center の詳細設定または SQL コマンドで行う。また AI mode 有効時の制限事項（VARIANT/OBJECT/ARRAY 列は対象外、300 列超のテーブルは対象外、クロスリージョン推論の有効化が必要な場合がある）と、LLM 使用に伴う トークン課金 が発生する旨が追記された。 影響: Enterprise Edition 以上のユーザーは AI mode を有効化することで分類精度を向上できるが、従来のコンピュート課金に加えトークン課金が追加発生するためコスト見積もりの見直しが必要となる。
変更内容: 本文を更新（25行追加・1行削除）
--- ahttps://docs.snowflake.com/en/user-guide/classify-intro+++ bhttps://docs.snowflake.com/en/user-guide/classify-intro@@ -70,4 +70,18 @@ You can also [use SQL commands](/user-guide/classify-auto) to create and modify a classification profile. If you are using SQL, associating the classification profile with a database to start the classification process is a separate step.++<a id="label-classify-ai-mode"></a>++### AI mode++Available to Enterprise Edition or higher.++A classification profile can optionally use **AI mode** to use LLMs to identify additional semantic categories beyond those identified by+standard classification. When AI mode is enabled, Snowflake uses the `openai-gpt-5-mini` model to augment classification results and improve+accuracy.++You can enable AI mode when you [set up classification with advanced settings](/user-guide/classify-ui-trust-center#label-classify-trust-center-advanced)+in the Trust Center, or by using [SQL commands](/user-guide/classify-auto#label-classify-auto-ai-mode) to create or update a classification+profile. ## Protecting sensitive data@@ -121,5 +135,9 @@ Sensitive data classification consumes credits as it uses [serverless compute resources](#label-serverless-credit-usage) to-classify tables in the database. For more information about pricing for this consumption, see Table 5 in the+classify tables in the database.++When you enable [AI mode](#label-classify-ai-mode) on a classification profile, you also incur token charges for LLM usage in addition to these compute charges.++For more information about pricing for this consumption, see Table 5 in the [Snowflake Service Consumption Table](https://www.snowflake.com/legal-files/CreditConsumptionTable.pdf). @@ -245,2 +263,8 @@ - A column with a name that has more than 255 characters. - A column with a name that includes the `$` character.+- When [AI mode](#label-classify-ai-mode) is enabled on a classification profile:+ - Columns of type VARIANT, OBJECT, or ARRAY are not evaluated in AI mode classification.+ - Tables with more than 300 columns are not passed for AI enrichment.+ - AI mode is only available with the OpenAI GPT-5 Mini model (`openai-gpt-5-mini`).+ - To use AI mode, you might need to [enable cross-region inference](/user-guide/snowflake-cortex/cross-region-inference)+ on your account.
background パラメータの説明が改訂され、stream パラメータとの組み合わせによる動作が明確化された。従来は非同期実行時の挙動のみ記載されていたが、改訂後は stream=false の場合は即座に status: in_progress と run_id を返す動作、stream=true の場合は Server-Sent Events でストリーミングされる動作がそれぞれ区別して記載された。
影響: background=true かつ stream=true でSSEストリーミングが利用できることがドキュメント上明示されたため、非同期バックグラウンド実行でもリアルタイムにレスポンスを受け取る実装が可能であることが確認できる。
変更内容: 本文を更新（9行追加・9行削除）
--- ahttps://docs.snowflake.com/en/sql-reference/functions/data_agent_run-snowflake-cortex+++ bhttps://docs.snowflake.com/en/sql-reference/functions/data_agent_run-snowflake-cortex@@ -115,13 +115,13 @@ <td> Whether to run the agent asynchronously. If `true`, the agent runs- asynchronously and returns immediately with `"status": "in_progress"`- and a `run_id`. The run continues in the background with a 6 hour- timeout, even if the client disconnects. Retrieve the response after- the run completes by streaming from the Stream Agent Run endpoint with- the `run_id` (REST API) or by polling- [](/sql-reference/functions/thread_messages-snowflake-cortex) with the- thread ID (SQL function). If `false`, the agent runs synchronously with- a 15 minute timeout. Only available when using threads to manage- conversation history.+ asynchronously in the background with a 6 hour timeout, even if the+ client disconnects. For a background run with `stream` set to `false`,+ the API returns immediately with `status: in_progress` and a `run_id`;+ retrieve the response after the run completes by streaming from the+ Stream Agent Run endpoint with the `run_id` (REST API) or by polling the+ THREAD_MESSAGES SQL function with the thread ID. If `stream` is `true`,+ the response is streamed as Server-Sent Events. If `background` is+ `false`, the agent runs synchronously with a 15 minute timeout. Only+ available when using threads to manage conversation history. </td> </tr>
AGENT_RUN 関数の background パラメータの説明が書き直された。従来は非同期実行時の動作を一括で記述していたが、改訂後は stream が false の場合（即座に status: in_progress と run_id を返す）と stream が true の場合（Server-Sent Events でストリーミング）を明確に分けて説明するようになった。機能の追加・削除はなく、既存の動作をより正確に記述した文書改善。
影響: background=true かつ stream=true でSSEストリーミングされる動作が明文化されたため、非同期ストリーミングを利用する際のAPI設計の参考になる。
変更内容: 本文を更新（9行追加・9行削除）
--- ahttps://docs.snowflake.com/en/sql-reference/functions/agent_run-snowflake-cortex+++ bhttps://docs.snowflake.com/en/sql-reference/functions/agent_run-snowflake-cortex@@ -82,13 +82,13 @@ <td> Whether to run the agent asynchronously. If `true`, the agent runs- asynchronously and returns immediately with `"status": "in_progress"`- and a `run_id`. The run continues in the background with a 6 hour- timeout, even if the client disconnects. Retrieve the response after- the run completes by streaming from the Stream Agent Run endpoint with- the `run_id` (REST API) or by polling- [](/sql-reference/functions/thread_messages-snowflake-cortex) with the- thread ID (SQL function). If `false`, the agent runs synchronously with- a 15 minute timeout. Only available when using threads to manage- conversation history.+ asynchronously in the background with a 6 hour timeout, even if the+ client disconnects. For a background run with `stream` set to `false`,+ the API returns immediately with `status: in_progress` and a `run_id`;+ retrieve the response after the run completes by streaming from the+ Stream Agent Run endpoint with the `run_id` (REST API) or by polling the+ THREAD_MESSAGES SQL function with the thread ID. If `stream` is `true`,+ the response is streamed as Server-Sent Events. If `background` is+ `false`, the agent runs synchronously with a 15 minute timeout. Only+ available when using threads to manage conversation history. </td> </tr>
Query Insights のインサイト名 QUERY_INSIGHT_OPTIMA_PLANNING_USED が QUERY_INSIGHT_SNOWFLAKE_OPTIMA_PLANNING_USED にリネームされ、新たに QUERY_INSIGHT_SNOWFLAKE_OPTIMA_PLANNING_OPPORTUNITY が追加された。後者は Optima Planning で改善可能だが、対応していないウェアハウス世代で実行されたクエリに表示され、Generation 2 Standard Warehouses または Adaptive Warehouses への移行を案内する。
影響: Optima Planning 非対応のウェアハウスで実行中のクエリに新しいインサイトが表示されるようになるため、ウェアハウスの世代移行を検討する判断材料として活用できる。
変更内容: 本文を更新（14行追加・1行削除）
--- ahttps://docs.snowflake.com/en/user-guide/query-insights+++ bhttps://docs.snowflake.com/en/user-guide/query-insights@@ -37,5 +37,6 @@ | `QUERY_INSIGHT_SEARCH_OPTIMIZATION_USED` | [](#label-query-insight-search-optimization-used) | | `QUERY_INSIGHT_SNOWFLAKE_OPTIMA` | [](#label-query-insight-snowflake-optima) |-| `QUERY_INSIGHT_OPTIMA_PLANNING_USED` | [](#label-query-insight-optima-planning-used) |+| `QUERY_INSIGHT_SNOWFLAKE_OPTIMA_PLANNING_USED` | [](#label-query-insight-optima-planning-used) |+| `QUERY_INSIGHT_SNOWFLAKE_OPTIMA_PLANNING_OPPORTUNITY` | [](#label-query-insight-snowflake-optima-planning-opportunity) | | `QUERY_INSIGHT_SEARCH_OPTIMIZATION_AND_SNOWFLAKE_OPTIMA` | [](#label-query-insight-search-optimization-and-snowflake-optima) | | `QUERY_INSIGHT_JOIN_WITH_NO_JOIN_CONDITION` | [](#label-query-insight-join-with-no-join-condition) |@@ -112,4 +113,16 @@ The query benefited from [Optima Planning](/user-guide/snowflake-optima#optima-planning). Snowflake applied a learned execution plan that improved performance for this query.++<a id="label-query-insight-snowflake-optima-planning-opportunity"></a>++### Query could benefit from Snowflake Optima Planning++Optima Planning can improve this query, but the query ran on a warehouse generation that doesn't support+[Optima Planning](/user-guide/snowflake-optima#optima-planning), so Snowflake didn't apply a learned execution+plan.++Optima Planning is available on [Snowflake generation 2 standard warehouses](/user-guide/warehouses-gen2) and+[Adaptive Warehouses](/user-guide/warehouses-adaptive). To let Snowflake apply a learned execution plan that can+improve performance for this query, run it on one of these warehouse types. <a id="label-query-insight-search-optimization-and-snowflake-optima"></a>
CoCo CLI にログ関連の環境変数が3つ追加された。COCO_LOG_LEVEL（ログレベル制御、デフォルト INFO）、COCO_DEBUG（デバッグログ強制有効化）、COCO_LOG_DIR（ログ出力先ディレクトリ指定）。また、ディレクトリ構造の説明で logs/ フォルダ内のファイル名（coco.log, coco_errors.log）が明記された。
影響: トラブルシューティング時に COCO_DEBUG や COCO_LOG_LEVEL を設定することで、ログの詳細度や出力先を制御できるようになる。
変更内容: 本文を更新（13行追加・1行削除）
--- ahttps://docs.snowflake.com/en/user-guide/cortex-code/settings+++ bhttps://docs.snowflake.com/en/user-guide/cortex-code/settings@@ -64,5 +64,5 @@ │ ├── table_cache.json # SQL result metadata │ └── sql_result_cache/ # Parquet files-├── logs/ # Log files+├── logs/ # Log files (coco.log, coco_errors.log) ├── memory/ # Persistent memory ├── agents/ # Custom agents@@ -177,4 +177,16 @@ <td>Requires confirmation for SQL write operations in bypass mode.</td> </tr>+ <tr>+ <td>`COCO_LOG_LEVEL`</td>+ <td>Minimum log level written to disk: `DEBUG`, `INFO`, `WARNING`, or `ERROR`. Default: `INFO`.</td>+ </tr>+ <tr>+ <td>`COCO_DEBUG`</td>+ <td>Set to `1`, `true`, or `yes` to force debug logging. Overrides `COCO_LOG_LEVEL`.</td>+ </tr>+ <tr>+ <td>`COCO_LOG_DIR`</td>+ <td>Directory for `coco.log` and `coco_errors.log`. Default: `$SNOWFLAKE_HOME/cortex/logs`, where `SNOWFLAKE_HOME` defaults to `~/.snowflake`.</td>+ </tr> </tbody>
Trust Center での機密データ分類設定手順に、新しいステップ 14 として AI Mode オプションが追加された。AI Mode を有効にすると、大規模言語モデル（現時点では OpenAI GPT-5 Mini のみ対応） を使用して分類精度を向上させることができる。これに伴い、既存の手順番号が 14〜16 から 15〜17 に繰り下がった。 影響: 分類精度を高めたい場合は AI Mode を有効化することで LLM ベースの分類が利用可能になるが、現時点でサポートされるモデルは GPT-5 Mini のみである点に留意が必要。
変更内容: 本文を更新（9行追加・3行削除）
--- ahttps://docs.snowflake.com/en/user-guide/classify-ui-trust-center+++ bhttps://docs.snowflake.com/en/user-guide/classify-ui-trust-center@@ -92,7 +92,13 @@ 13. Specify the database, schema, and name of the [classification profile](#label-classify-classification-profiles) where all of your settings will be saved.-14. Select the cadence at which previously classified objects are re-classified.-15. Specify if you want to exclude certain objects from the classification process. For information about excluding specific objects, see [](/user-guide/classify-auto-exclude).-16. Select **Enable**.+14. In the **AI Mode** section, optionally select **AI Mode** to improve the accuracy of classification using a large language model.+++ Currently, AI mode only supports the OpenAI GPT-5 Mini model.+++15. Select the cadence at which previously classified objects are re-classified.+16. Specify if you want to exclude certain objects from the classification process. For information about excluding specific objects, see [](/user-guide/classify-auto-exclude).+17. Select **Enable**. <a id="label-classify-trust-center-review-results"></a>
ノートブックをパッケージにバンドルする際、従来の方法（application_content.notebooks や単純なパッケージ追加）ではなく、workspace ディレクトリに配置しマニフェストで宣言する方式に変更された。具体的には、ノートブック作成後の手順が「パッケージに追加」から「workspace ディレクトリに追加」に修正され、workspace に関するドキュメントへのリンクが追加された。また、注意事項として application_content.notebooks ではなく workspace ディレクトリでノートブックを共有するよう明記された。
影響: Declarative Share でノートブックをバンドルする場合、workspace ディレクトリを使用する方式への移行が必要となり、既存の `application_content.notebooks` を使った共有方法は非推奨となる。
変更内容: 本文を更新（7行追加・4行削除）
--- ahttps://docs.snowflake.com/en/developer-guide/declarative-sharing/tutorials/create-share-coco+++ bhttps://docs.snowflake.com/en/developer-guide/declarative-sharing/tutorials/create-share-coco@@ -121,6 +121,7 @@ editor can corrupt notebook JSON, so don't ask CoCo Web to write `.ipynb` files. To bundle a notebook with your package from CoCo Web,-create the notebook directly in Snowsight and then add it to your-package.+create the notebook directly in Snowsight and then add it to a+workspace directory in your package. For more information, see+[](/developer-guide/declarative-sharing/workspaces). <a id="label-coco-tutorial-run-the-prompt"></a>@@ -416,4 +417,6 @@ [](/developer-guide/declarative-sharing/workspaces). +- Share notebooks in a workspace directory, not with `application_content.notebooks`. For more+ information, see [](/developer-guide/declarative-sharing/workspaces). - Notebooks can only access data inside the same application package. - Ensure each code cell has correct `language` metadata; otherwise SQL cells@@ -460,6 +463,6 @@ - **Add a Cortex Agent over a shared semantic view** — `Add a Cortex Agent over .<SEMANTIC_VIEW> to my . Put it in its own agent schema, update the manifest, and release LIVE.`-- **Bundle a getting-started notebook** (CoCo CLI only) —- `Create a notebook with three sample queries against the shared views in , add it to the package, and re-release LIVE.`+- **Bundle a getting-started notebook in a workspace** (CoCo CLI only) —+ `Create a notebook with three sample queries against the shared views in , put it in a workspace directory, declare the workspace in the manifest, and re-release LIVE.` <a id="label-coco-tutorial-cleanup-optional"></a>
プレビュー機能一覧に AI mode for sensitive data classification が新規追加された。2026年8月からオープンプレビューとして利用可能で、分類プロファイルでAIモードを有効にすると、LLMを使って標準分類では検出できない追加のセマンティックカテゴリを識別できる。Trust CenterまたはSQLから設定可能。 影響: 機密データ分類の精度向上を検討しているユーザーは、AIモードをプレビューで試用し、標準分類では捕捉できないカテゴリの検出可否を評価できる。
変更内容: 本文を更新（7行追加・0行削除）
--- ahttps://docs.snowflake.com/en/release-notes/preview-features+++ bhttps://docs.snowflake.com/en/release-notes/preview-features@@ -149,4 +149,11 @@ </tr> <tr>+ <td>AI mode for sensitive data classification</td>+ <td>Open</td>+ <td>August 2026</td>+ <td>[](/user-guide/classify-intro#label-classify-ai-mode)</td>+ <td>Use LLMs to identify additional semantic categories beyond standard classification when you enable AI mode on a classification profile in the Trust Center or with SQL.</td>+ </tr>+ <tr> <td>Metadata-based sensitive data classification recommendations in Snowsight</td> <td>Open</td>
warehouse パラメータの説明文が更新された。句読点の微修正（ピリオド→セミコロン）に加え、未指定時はユーザーのデフォルトウェアハウスが使用されるという動作説明が追加された。 影響: warehouse パラメータを省略してもデフォルトウェアハウスで動作することが明文化されたため、API呼び出し時にウェアハウス指定を省略する運用が公式にサポートされていると確認できる。
変更内容: 本文を更新（3行追加・2行削除）
--- ahttps://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-rest-api+++ bhttps://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-rest-api@@ -985,6 +985,7 @@ <td>string</td> <td>- The name of the warehouse. Case-sensitive, if it is an unquoted- identifier, provide the name in all-caps.+ The name of the warehouse. Case-sensitive; if it is an unquoted+ identifier, provide the name in all-caps. If not specified, the+ user's default warehouse is used. </td> </tr>
Notebooks in Workspaces のサービスに関する制限事項が更新された。ワークスペース間でのサービス共有不可・同一ワークスペース内でのデフォルト共有という従来の記述が削除され、代わりにサービスはワークスペースをまたいで共有可能だがユーザー単位でスコープされるという新しい仕様に変更された。また、x86 ベースの compute pool のみサポートされ、ARM ベースは未対応という制限が追加された。 影響: ARM ベースの compute pool を利用中のユーザーはNotebook サービスでは使用できないため、x86 ベースへの切り替えが必要となる。また、サービス共有の範囲が変わったため、既存のワークスペース運用を見直す必要がある。
変更内容: 本文を更新（2行追加・3行削除）
--- ahttps://docs.snowflake.com/en/user-guide/ui-snowsight/notebooks-in-workspaces/notebooks-in-workspaces-limitations+++ bhttps://docs.snowflake.com/en/user-guide/ui-snowsight/notebooks-in-workspaces/notebooks-in-workspaces-limitations@@ -19,7 +19,6 @@ - The number of notebook services allowed in an account is subject to the Snowpark Container Services [account limits](/developer-guide/snowpark-container-services/working-with-compute-pool#guidelines-and-limitations).-- Notebooks in different workspaces cannot share a service.-- Notebooks in the same workspace connect to a shared service by default.-- Users may create multiple services within a workspace and assign notebooks to different services as needed.+- Notebook services can only run on x86-based compute pools. ARM-based compute pools are not yet supported.+- Users may create multiple services and assign notebooks to different services as needed. Services can be shared across Workspaces but are scoped to per user. - Notebook services may be restarted over the weekend for container service maintenance. After a restart, you must rerun notebooks and reinstall any packages to restore variables and packages. For more information, see [Service maintenance](#label-nb-in-ws-service-maintenance).
IS_ROLE_ACTIVATED 関数の引数に関する説明が追加された。複数のロール名を指定でき、いずれかがアクティブであれば 'TRUE' を返す仕様が明記された。ただし複数ロール指定は各ロールが定数値（文字列リテラル等）の場合のみ可能で、定数でない式の場合は単一ロールのみ指定可能という制約も記載された。 影響: IS_ROLE_ACTIVATED で複数ロールを一括チェックできることが公式に文書化されたため、ロール判定ロジックの簡略化に活用できる。
変更内容: 本文を更新（4行追加・0行削除）
--- ahttps://docs.snowflake.com/en/sql-reference/functions/sys_context_snowflake_session+++ bhttps://docs.snowflake.com/en/sql-reference/functions/sys_context_snowflake_session@@ -264,4 +264,8 @@ Arguments to pass to the function that you want to call. +[IS_ROLE_ACTIVATED](/sql-reference/functions/is_role_activated) accepts one or more role names and returns `'TRUE'` if any of+the roles is activated. You can specify multiple roles only when each role is a constant value, such as a string literal. When+the role is an expression that isn't a constant, you can specify only a single role.+ </dd> </dl>
CoCo CLI の利用要件に関する注記が追加された。標準の Snowflake トライアルアカウント（trial.snowflake.com で作成したもの）では CoCo CLI を利用できないことが明記され、利用には有料アカウントまたは専用の CoCo CLI トライアルが必要と説明された。 影響: 標準トライアルアカウントで CoCo CLI を試そうとしていたユーザーは、専用の CoCo CLI トライアルへのサインアップが必要となる。
変更内容: 本文を更新（3行追加・1行削除）
--- ahttps://docs.snowflake.com/en/user-guide/cortex-code/cortex-code-cli+++ bhttps://docs.snowflake.com/en/user-guide/cortex-code/cortex-code-cli@@ -12,5 +12,7 @@ Before you begin, ensure you have a Snowflake account with access to the required Cortex models. See [](#label-cortex-code-cli-prerequisites) for full details. -If you do not have a Snowflake account, you can [sign up for a free CoCo CLI trial](https://signup.snowflake.com/cortex-code).+If you don't have a Snowflake account, you can [sign up for a free CoCo CLI trial](https://signup.snowflake.com/cortex-code).++CoCo CLI is not available on standard Snowflake trial accounts (for example, accounts created at [trial.snowflake.com](https://trial.snowflake.com)). To use CoCo CLI, you need either a paid Snowflake account or the dedicated CoCo CLI trial linked above. ## Install CoCo CLI
SYS_CONTEXT 関数の引数説明セクションに、IS_ROLE_ACTIVATED 関数の動作詳細が追加された。IS_ROLE_ACTIVATED は1つ以上のロール名を受け取り、いずれかがアクティブであれば 'TRUE' を返す。複数ロールを指定できるのは各ロールが定数値（文字列リテラル等）の場合のみで、定数でない式の場合は単一ロールのみ指定可能という制約が明記された。
影響: SYS_CONTEXT 経由で IS_ROLE_ACTIVATED を使う際、複数ロール指定は定数値に限られるという制約を意識する必要がある。
変更内容: 本文を更新（4行追加・0行削除）
--- ahttps://docs.snowflake.com/en/sql-reference/functions/sys_context_snowflake_current+++ bhttps://docs.snowflake.com/en/sql-reference/functions/sys_context_snowflake_current@@ -104,4 +104,8 @@ Arguments to pass to the function that you want to call. +[IS_ROLE_ACTIVATED](/sql-reference/functions/is_role_activated) accepts one or more role names and returns `'TRUE'` if any of+the roles is activated. You can specify multiple roles only when each role is a constant value, such as a string literal. When+the role is an expression that isn't a constant, you can specify only a single role.+ </dd> </dl>
（この変更は要約対象外です。diff を参照してください）
変更内容: 3行以下の小規模な更新
--- ahttps://docs.snowflake.com/en/sql-reference/account-usage/query_insights+++ bhttps://docs.snowflake.com/en/sql-reference/account-usage/query_insights@@ -131,5 +131,6 @@ - `QUERY`: Insights about the efficiency of query execution plans. This label applies to the following types of insights:- - [QUERY_INSIGHT_OPTIMA_PLANNING_USED](#label-query-insight-optima-planning-used)+ - [QUERY_INSIGHT_SNOWFLAKE_OPTIMA_PLANNING_USED](#label-query-insight-optima-planning-used)+ - [QUERY_INSIGHT_SNOWFLAKE_OPTIMA_PLANNING_OPPORTUNITY](#label-query-insight-snowflake-optima-planning-opportunity) - `JOIN`: Insights about the efficiency of JOIN operations in the query. This label applies to the following types of insights:
（この変更は要約対象外です。diff を参照してください）
変更内容: 3行以下の小規模な更新
--- ahttps://docs.snowflake.com/en/sql-reference/classes/classification_profile+++ bhttps://docs.snowflake.com/en/sql-reference/classes/classification_profile@@ -19,4 +19,5 @@ - [<classification_profile_name>!DESCRIBE](/sql-reference/classes/classification_profile/methods/describe)+- [<classification_profile_name>!SET_AI_MODE](/sql-reference/classes/classification_profile/methods/set_ai_mode) - [<classification_profile_name>!SET_AUTO_TAG](/sql-reference/classes/classification_profile/methods/set_auto_tag) - [<classification_profile_name>!SET_CLASSIFY_VIEWS](/sql-reference/classes/classification_profile/methods/set_classify_views)
| セクション | S | A | B | C | 計 |
|---|---|---|---|---|---|
| SQL Functions | 0 | 1 | 4 | 0 | 5 |
| User Guide | 0 | 1 | 3 | 0 | 4 |
| Developer Guide | 0 | 2 | 1 | 0 | 3 |
| SQL Classes | 0 | 1 | 0 | 1 | 2 |
| Release Notes | 0 | 0 | 2 | 0 | 2 |
| Snowflake Cortex (AI & ML) | 0 | 0 | 2 | 0 | 2 |
| Cortex Code | 0 | 0 | 2 | 0 | 2 |
| Snowsight UI | 0 | 0 | 1 | 0 | 1 |
| Account Usage | 0 | 0 | 0 | 1 | 1 |