arfi 0.5.0 → 1.0.0

data/README.md

@@ -7,45 +7,52 @@
 
 ---
 
-> [!WARNING]
-> This project only supports PostgreSQL and MySQL databases. SQLite3 will be supported in the future as well as other
-> databases supported by Rails.
+ARFI – *ActiveRecord Functions Integration*
 
-> [!NOTE]
-> This project requires Ruby 3.1.0+, in future updated 2.6+ Ruby versions will be supported.
+ARFI helps Rails apps create and maintain **custom SQL functions** (often used for functional indexes and query helpers)
+without switching from `schema.rb` to `structure.sql`.
 
----
+ARFI follows a **"current state"** model: your repository contains the current function definitions in
+`db/functions/**`, and ARFI loads them into the database automatically during common Rails DB tasks
+(e.g., `db:prepare`, `db:test:prepare`, `db:migrate`, etc.).
+
+PostgreSQL bonus: ARFI can recover at runtime from `PG::UndefinedFunction` by loading managed function files and
+retrying the failed query once (thread-guarded).
+
+ARFI supports both single-DB and multi-DB Rails setups.
 
-ARFI – *ActiveRecord Functional Indexes*
+Demo project: https://github.com/unurgunite/poc_arfi_72
 
-The ARFI gem provides the ability to create and maintain custom SQL functions for ActiveRecord models without switching
-to `structure.sql` (an SQL-based schema). You can use your own SQL functions in any part of the project, from migrations
-and models to everything else. There is a working example in
-the [demo project](https://github.com/unurgunite/poc_arfi_72). All instructions are described
-in [README](https://github.com/unurgunite/poc_arfi_72/blob/master/README.md). ARFI supports all types of database
-architectures implemented in Rails, suitable for both working with single databases and for simultaneous work with
-multiple databases in the same environment.
+---
 
 * [ARFI](#arfi)
     * [Installation](#installation)
     * [Usage](#usage)
         * [Internal documentation](#internal-documentation)
         * [CLI](#cli)
-        * [Project creation](#project-creation)
-        * [Index creation](#index-creation)
-        * [Index destroy](#index-destroy)
+        * [Project initialization](#project-initialization)
+        * [Directory layout (1.0.0+)](#directory-layout-100)
+        * [Function creation](#function-creation)
+        * [Function destroy](#function-destroy)
         * [Additional help](#additional-help)
+    * [Architecture](#architecture)
+        * [Entry points](#entry-points)
+        * [Decision flow](#decision-flow)
+        * [Key design points](#key-design-points)
     * [Demo](#demo)
     * [Library features](#library-features)
-    * [Roadmap](#roadmap)
+    * [TODO](#todo)
     * [Commands](#commands)
-        * [Function creation](#function-creation)
-        * [Function destroy](#function-destroy)
+        * [Function creation](#function-creation-1)
+        * [Function destroy](#function-destroy-1)
             * [Options](#options)
                 * [`--template` option](#--template-option)
                 * [`--adapter` option](#--adapter-option)
+                * [`--schema` option](#--schema-option)
+                * [`--force` option](#--force-option)
     * [Limitations](#limitations)
     * [Development](#development)
+        * [Run tests locally](#run-tests-locally)
         * [Build from source](#build-from-source)
     * [Requirements](#requirements)
     * [Contributing](#contributing)
@@ -71,28 +78,161 @@ Internal documentation available at https://github.com/unurgunite/arfi_docs.
 
 ARFI uses Thor as a command line interface (CLI) instead of Rake, so it has a specific DSL.
 
-### Project creation
+### Project initialization
+
+Run:
+
+```bash
+bundle exec arfi init
+```
+
+This ensures the directory structure exists:
+
+```
+db/functions/public
+db/functions/postgresql/public
+db/functions/mysql/public
+```
+
+Backwards-compatible alias:
+
+```bash
+bundle exec arfi project create
+```
+
+### Directory layout (1.0.0+)
+
+Canonical layout (explicit `public`):
+
+- Generic public: `db/functions/public/<function>.sql`
+- PostgreSQL public: `db/functions/postgresql/public/<function>.sql`
+- PostgreSQL schema: `db/functions/postgresql/<schema>/<function>.sql`
+- MySQL / MariaDB / Trilogy public: `db/functions/mysql/public/<function>.sql`
 
-Firstly, run `bundle exec arfi project create` to create a new project. This command will create `db/functions`
-directory. ARFI uses `db/functions` directory to store your SQL functions.
+Adapter-specific files override generic files by basename (same `<function>.sql`).
+Files starting with `_` are ignored.
 
 ### Function creation
 
-Run `bundle exec arfi f_idx create function_name` to create a new function. New SQL function will be created in
-`db/functions` directory under `function_name_v01.sql` name. Edit your function and run `bundle exec rails db:migrate`.
-You can also use custom template for functions using `--template` flag, this behaviour is described below.
-Type `bundle exec arfi f_idx help create` for additional info.
+Run:
+
+```bash
+bundle exec arfi functions create function_name
+```
+
+The file location depends on adapter:
+
+- PostgreSQL: `db/functions/postgresql/public/function_name.sql`
+- MySQL/Trilogy: `db/functions/mysql/public/function_name.sql`
+- Generic: `db/functions/public/function_name.sql`
+
+Edit the function SQL and run your usual DB task (`db:migrate`, `db:prepare`, etc.).
+
+You can also use a custom template for functions using the `--template` flag; this behaviour is described below.
+Type `bundle exec arfi functions help create` for additional info.
+
+Backwards-compatible alias:
+
+```bash
+bundle exec arfi f_idx create function_name
+```
 
 ### Function destroy
 
-If you want to destroy your function, run `bundle exec arfi f_idx destroy function_name [revision (default 1)]`. Please
-note that after deleting the function, it will still be available, but if you run "bundle exec rails db:migrate" again,
-an error will occur when using the function. Enter `bundle exec arfi f_idx help destroy` for more information.
+Destroy deletes the function file from disk:
+
+```bash
+bundle exec arfi functions destroy function_name
+```
+
+Backwards-compatible alias:
+
+```bash
+bundle exec arfi f_idx destroy function_name
+```
 
 ### Additional help
 
 Run `bundle exec arfi` for additional help.
 
+## Architecture
+
+ARFI loads SQL function files into your database(s) during common Rails DB tasks. The flow depends on the entry point, database configuration, and adapter.
+
+### Entry points
+
+| Entry point                 | Description                                        |
+|-----------------------------|----------------------------------------------------|
+| `db:migrate` / `db:prepare` | Non-suffixed Rake tasks -> `_db:arfi_enhance`      |
+| `db:migrate:animals`        | Suffixed Rake task -> `run_with_connection_switch` |
+| Runtime retry               | `PG::UndefinedFunction` -> auto-reload             |
+| Direct call                 | `SqlFunctionLoader.load!` from anywhere            |
+
+### Decision flow
+
+```mermaid
+flowchart TD
+    A(["SqlFunctionLoader.load!"]) --> B{"connection<br>passed?"}
+    B -->|"Yes"| C["Use passed connection"]
+    B -->|"No"| D{"multi_db?<br>AND<br>task_name nil?"}
+
+    D -->|"No"| E["populate_db<br>(default_connection)"]
+    D -->|"Yes"| F["populate_multiple_db"]
+
+    F --> F1["Save original connection config"]
+    F1 --> F2["For each DB config:"]
+    F2 --> F3["establish_connection(config)"]
+    F3 --> F4["populate_db(conn)"]
+    F4 --> F2
+    F2 --> F5["Restore original connection config"]
+
+    C --> G["Check adapter support"]
+    E --> G
+    F5 --> G
+
+    G --> H["Collect SQL files"]
+    H --> I["db/functions/public/*.sql<br>(generic, priority 1-2)"]
+    H --> J["db/functions/{pg,mysql}/**/*.sql<br>(adapter-specific, priority 8-10)"]
+
+    I --> K["Override resolution:<br>higher priority wins"]
+    J --> K
+    K --> L["Ignore _prefixed files"]
+    L --> M["Execute each SQL file"]
+
+    M --> N{"clear_active_connections?"}
+    N -->|"Yes"| O["connection_handler<br>.clear_active_connections!"]
+    N -->|"No"| P["Skip cleanup"]
+
+    subgraph Entry["Entry Points"]
+        Q1["Rake: db:migrate"]
+        Q2["Rake: db:migrate:animals"]
+        Q3["Runtime: PG::UndefinedFunction"]
+        Q4["Direct call"]
+    end
+
+    Q1 --> R1["_db:arfi_enhance<br>load!(task_name: nil)"]
+    R1 --> A
+
+    Q2 --> R2["_db:arfi_enhance:db:migrate:animals"]
+    R2 --> R3["run_with_connection_switch"]
+    R3 --> R4["Save original config"]
+    R4 --> R5["establish_connection(animals)"]
+    R5 --> A
+    R5 --> R6["Restore original config"]
+
+    Q3 --> R7["arfi_try_reload_and_retry?"]
+    R7 --> R8["load!(task_name: 'arfi:runtime',<br>connection: self,<br>clear_active_connections: false)"]
+    R8 --> A
+```
+
+### Key design points
+
+- **Single-DB (default)**: loads functions into one connection, one `populate_db` call.
+- **Multi-DB**: `populate_multiple_db` iterates over all configs for the current environment, loading functions into each. The original connection is saved before the loop and restored afterwards, matching the pattern used in `run_with_connection_switch`.
+- **Suffixed tasks** (`db:migrate:animals`): switch to the named database, load functions, then restore the original connection.
+- **Runtime retry**: triggered only for `PG::UndefinedFunction`, loads functions on the same connection without clearing other active connections.
+- **File override**: adapter-specific files take priority over generic files. The highest priority file wins per schema+filename key.
+
 ## Demo
 
 Demo available as separate project built with Rails 7.2 and PostgreSQL 14: https://github.com/unurgunite/poc_arfi_72.
@@ -100,43 +240,42 @@ README is also available.
 
 ## Library features
 
-1. ARFI supports about all types of database initialization. It respects your database schema format and database
-   configuration.
-
-   | Task             | Completed                                                    |
-   |------------------|--------------------------------------------------------------|
-   | db:migrate       | :white_check_mark:                                           |
-   | db:setup         | :white_check_mark:                                           |
-   | db:prepare       | :white_check_mark:                                           |
-   | db:schema:load   | :white_check_mark:                                           |
-   | db:reset         | :white_check_mark:                                           |
-   | db:setup:db_name | In progress (see [limitations][1]) :arrows_counterclockwise: |
-
-2. Database support. ARFI supports PostgreSQL and MySQL databases and projects with multiple databases at the same time.
-
-   | DB adapter | Tested                                |
-   |------------|---------------------------------------|
-   | PostgreSQL | :white_check_mark:                    |
-   | MySQL      | :white_check_mark:                    |
-   | SQLite3    | In progress :arrows_counterclockwise: |
+1. ARFI supports most Rails database initialization flows and respects your schema format and database configuration.
+
+   | Task                     | Supported |
+   |--------------------------|-----------|
+   | db:migrate               | ✅         |
+   | db:setup                 | ✅         |
+   | db:prepare               | ✅         |
+   | db:test:prepare          | ✅         |
+   | db:schema:load           | ✅         |
+   | db:migrate:db_name       | ✅         |
+   | db:prepare:db_name       | ✅         |
+   | db:schema:load:db_name   | ✅         |
+
+2. Database support. ARFI supports PostgreSQL and MySQL-compatible databases, including multi-db setups.
+
+   | DB adapter / client   | Tested                          |
+   |-----------------------|---------------------------------|
+   | PostgreSQL            | ✅                               |
+   | MySQL (mysql2)        | ✅                               |
+   | MariaDB (mysql2)      | ✅                               |
+   | Trilogy (Rails 7.1+)  | ✅                               |
+   | SQLite3               | Not supported (see Limitations) |
 
 3. Rails support
 
-   | Rails version | Tested                                |
-   |---------------|---------------------------------------|
-   | 8             | :white_check_mark:                    |
-   | 7             | :white_check_mark:                    |
-   | 6             | In progress :arrows_counterclockwise: |
+   | Rails version | Tested |
+   |---------------|--------|
+   | 8             | ✅      |
+   | 7             | ✅      |
+   | 6             | ✅      |
 
-## Roadmap
+## TODO
 
-1. ~~Custom template for SQL functions using `--template` flag;~~
-2. ~~Multidb support (Rails 6+ feature);~~
-3. Add support for 4+ ActiveRecord;
-4. Add RSpec tests;
-5. ~~Add separate YARD doc page;~~
-6. ~~Update CI/CD;~~
-7. Add support for Ruby 2.6+.
+1. Add `functions validate` / `functions doctor` command group (planned for 1.1.0);
+2. Add functions autoloader (v1.2.0);
+3. Add more adapters (Oracle, MSSQL).
 
 ## Commands
 
@@ -145,79 +284,150 @@ ARFI uses Thor as a command line interface.
 
 ### Function creation
 
-ARFI supports creation of SQL functions. To create a new function, run `bundle exec arfi f_idx create function_name`.
+To create a new function file, run:
+
+```bash
+bundle exec arfi functions create function_name
+```
+
 Also, there are some options:
 
-| Option name  | Description                                                                          | Possible values            | Default value                                                 |
-|--------------|--------------------------------------------------------------------------------------|----------------------------|---------------------------------------------------------------|
-| `--template` | use custom template                                                                  | path within you filesystem | nil (will be used default template for each type of adapters) |
-| `--adapter`  | adapter specific function creation due to syntax differences between different RDBMS | postgresql, mysql          | nil (function will be stored in generic `db/functions`)       |
+| Option name  | Description                                                           | Possible values               | Default value                       |
+|--------------|-----------------------------------------------------------------------|-------------------------------|-------------------------------------|
+| `--template` | Use custom Ruby template that returns SQL                             | path within your filesystem   | nil                                 |
+| `--adapter`  | Store function in adapter directory and use adapter skeleton/template | postgresql, mysql, trilogy    | nil (generic `db/functions/public`) |
+| `--schema`   | PostgreSQL schema (alternative to `schema.function` form)             | schema name (postgresql only) | public                              |
+| `--force`    | Overwrite existing function file if it already exists                 | true/false                    | false                               |
+
+Backwards-compatible alias:
+
+```bash
+bundle exec arfi f_idx create function_name
+```
 
 ### Function destroy
 
-ARFI supports destroy of SQL functions. To destroy a function, run
-`bundle exec arfi f_idx destroy function_name [revision (1 by default)]`.
+To destroy a function file, run:
 
-| Option name  | Description                  | Possible values   | Default value                                              |
-|--------------|------------------------------|-------------------|------------------------------------------------------------|
-| `--revision` | Function revision to destroy | Integer           | 1                                                          |
-| `--adapter`  | adapter specific function    | postgresql, mysql | nil (function will be destroyed in generic `db/functions`) |
+```bash
+bundle exec arfi functions destroy function_name
+```
+
+| Option name | Description                                          | Possible values               | Default value |
+|-------------|------------------------------------------------------|-------------------------------|---------------|
+| `--adapter` | Adapter-specific function directory                  | postgresql, mysql, trilogy    | nil (generic) |
+| `--schema`  | PostgreSQL schema (alternative to `schema.function`) | schema name (postgresql only) | public        |
+
+Backwards-compatible alias:
+
+```bash
+bundle exec arfi f_idx destroy function_name
+```
 
 #### Options
 
 ##### `--template` option
 
 This option is used for creating an SQL function. In this case, the function will not be created with the default
-template, but with user defined. There are some rules for templates:
+template, but with a user-defined one. There are some rules for templates:
+
+1. The template must be written in Ruby-compatible syntax: the function must be placed in a HEREDOC statement and must
+   use interpolation for variables. You can use helper methods in the template file. The main rule is that the template
+   must evaluate to a String with SQL.
+2. ARFI supports dynamic variables in templates:
+    - `index_name` (backward compatible) / `function_name`
+    - `schema_name`
+    - `qualified_name`
+    - `original_ref`
 
-1. The template must be written in a Ruby-compatible syntax: the function must be placed in a HEREDOC statement and must
-   use interpolation for variables. If you need to take a more comprehensive approach to the issue of function
-   generation, you can try using your own methods in the template file. No matter what you write there, the main rule is
-   that your main method should return a string with a function template, as described below.
-2. ARFI supports dynamic variables in templates, but only one at the moment. You need to specify `index_name`
-   variable as below. In feature updated ARFI will support more variables. Here are default templates in ARFI for
-   PostgreSQL and MySQL:
+   Default templates:
 
    PostgreSQL:
-    ```ruby
-    <<~SQL
-      CREATE OR REPLACE FUNCTION #{index_name}() RETURNS TEXT[]
-      LANGUAGE SQL
-      IMMUTABLE AS
-      $$
-        -- Function body here
-      $$
-    SQL
-    ```
+   ```ruby
+   <<~SQL
+     CREATE OR REPLACE FUNCTION #{qualified_name}() RETURNS TEXT[]
+     LANGUAGE SQL
+     IMMUTABLE AS
+     $$
+       -- Function body here
+     $$
+   SQL
+   ```
+
    MySQL:
-    ```ruby
-    <<~SQL
-      CREATE FUNCTION #{index_name} ()
-      RETURNS return_type
-      BEGIN
-        -- Function body here
-      END;
-    SQL
-    ```
-3. By default ARFI uses PostgreSQL template.
+   ```ruby
+   <<~SQL
+     CREATE FUNCTION #{function_name} ()
+     RETURNS return_type
+     BEGIN
+       -- Function body here
+     END;
+   SQL
+   ```
+
+3. By default, ARFI uses PostgreSQL template.
 
 ##### `--adapter` option
 
 This option is used both when destroying and when creating an SQL function. In this case, the function will not be
-created in the default directory `db/functions`, but in the child `db/functions/#{adapter}`. Supported adapters:
-`postgresql`and `mysql`, but there will be more in the future.
+created in the default directory `db/functions/public`, but in an adapter directory.
+
+Supported adapters: `postgresql`, `mysql`, `trilogy`.
+
+##### `--schema` option
+
+PostgreSQL-only option that controls schema directory:
+
+- `--adapter=postgresql --schema=audit` => `db/functions/postgresql/audit/<fn>.sql`
+
+You can also pass schema-qualified names: `audit.my_fn`.
+
+##### `--force` option
+
+Overwrite existing function file if it already exists.
 
 ## Limitations
 
-Currently, ARFI has a limitation for `db:setup:db_name` task due to the fact how Rails manage this rake task. More info
-here: [limitations][1]. This command will work, but it is not recommended to use it. Note that this limitation applies
-only to multi-db setup, default `db:setup` will work as expected.
+- SQLite3 is not supported for "stored functions from SQL files" because SQLite user-defined functions are typically
+  registered per connection via the client library, not created via `CREATE FUNCTION` SQL.
+- MySQL does not support `CREATE OR REPLACE FUNCTION` in the same way as PostgreSQL; plan function replacement strategy
+  accordingly.
 
 ## Development
 
+### Run tests locally
+
+Start the required databases via Docker Compose:
+
+```shell
+docker compose up -d
+```
+
+Wait for both services to become healthy (check with `docker ps`). Then run the test suite:
+
+```shell
+bundle exec rspec
+```
+
+The test suite automatically detects available databases via environment variables:
+
+| Variable            | Default value                                             |
+|---------------------|-----------------------------------------------------------|
+| `ARFI_POSTGRES_URL` | `postgresql://postgres:postgres@localhost:5432/arfi_test` |
+| `ARFI_MYSQL_URL`    | `mysql2://root:password@127.0.0.1:3306/arfi_test`         |
+
+Specs tagged with `:pgsql` or `:mysql` only run when the corresponding database is reachable;
+untagged specs run with no external dependencies.
+
+To stop the containers when done:
+
+```shell
+docker compose down
+```
+
 ### Build from source
 
-The manual installation includes installation via command line interface. it is practically no different from what
+The manual installation includes installation via command line interface. It is practically no different from what
 happens during the automatic build of the project:
 
 ```shell
@@ -225,7 +435,7 @@ git clone https://github.com/unurgunite/arfi.git
 cd arfi
 bundle install
 gem build arfi.gemspec
-gem install arfi-0.5.0.gem
+gem install arfi-1.0.0.gem
 ```
 
 Also, you can run `bin/setup` to automatically install everything needed.
@@ -236,14 +446,14 @@ ARFI is built on top of the following gems:
 
 | Dependencies | Description                                                                                |
 |--------------|--------------------------------------------------------------------------------------------|
-| ActiveRecord | Used to patch `ActiveRecord::Base` module with new methods.                                |
+| ActiveRecord | Used to patch `ActiveRecord::Base` module with new methods                                 |
 | Rails        | Used for fetching project settings (database connection settings, Rails environment, etc.) |
-| Thor         | For CLI development.                                                                       |
-| Rubocop      | For static code analysis.                                                                  |
-| Rake         | For patching built-in Rails Rake tasks.                                                    |
-| Steep        | For static type checking.                                                                  |
-| RBS          | For static type checking.                                                                  |
-| YARD         | For generating documentation.                                                              |
+| Thor         | For CLI development                                                                        |
+| Rubocop      | For static code analysis                                                                   |
+| Rake         | For patching built-in Rails Rake tasks                                                     |
+| Steep        | For static type checking                                                                   |
+| RBS          | For static type checking                                                                   |
+| YARD         | For generating documentation                                                               |
 
 ## Contributing
 
@@ -253,7 +463,8 @@ the [code of conduct](https://github.com/unurgunite/arfi/blob/master/CODE_OF_CON
 
 ## Miscellaneous
 
-ARFI is highly inspired by https://github.com/teoljungberg/fx project.
+ARFI differs by focusing on keeping functions in sync from the current repository state (and optional runtime recovery
+on PostgreSQL).
 
 ## License
 
@@ -263,5 +474,3 @@ The gem is available as open source under the terms of the [MIT License](https:/
 
 Everyone interacting in the ARFI project's codebases, issue trackers, chat rooms and mailing lists is expected to follow
 the [code of conduct](https://github.com/unurgunite/arfi/blob/master/CODE_OF_CONDUCT.md).
-
-[1]: https://blog.saeloun.com/2021/10/27/rails-7-adds-database-specific-setup/#limitation

data/SECURITY.md

@@ -0,0 +1,17 @@
+# Security Policy
+
+## Reporting a Vulnerability
+
+If you discover a security vulnerability, please do **not** open a public issue.
+
+Instead, send a private report to the repository maintainers via GitHub's
+[private vulnerability reporting](https://github.com/unurgunite/arfi/security/advisories/new).
+
+You should receive a response within 48 hours. If you don't, please follow up
+to ensure the message was received.
+
+## Supported Versions
+
+| Version | Supported |
+|---------|-----------|
+| 1.x     | ✅         |

data/Steepfile

@@ -1,35 +1,7 @@
 # frozen_string_literal: true
 
-# D = Steep::Diagnostic
-#
 target :lib do
   signature 'sig'
-  # ignore_signature "sig/test"
-  # use "rbs_collection"
-  check 'lib' # Directory name
-  # check "path/to/source.rb"         # File name
-  # check "app/models/**/*.rb"        # Glob
-  # ignore "lib/templates/*.rb"
+  check 'lib'
   ignore 'lib/arfi/extensions/active_record/connection_adapters/postgresql/database_statements.rb'
-
-  # library "pathname"              # Standard libraries
-  # library "strong_json"           # Gems
-
-  # configure_code_diagnostics(D::Ruby.default)      # `default` diagnostics setting (applies by default)
-  # configure_code_diagnostics(D::Ruby.strict)       # `strict` diagnostics setting
-  # configure_code_diagnostics(D::Ruby.lenient)      # `lenient` diagnostics setting
-  # configure_code_diagnostics(D::Ruby.silent)       # `silent` diagnostics setting
-  # configure_code_diagnostics do |hash|             # You can setup everything yourself
-  #   hash[D::Ruby::NoMethod] = :information
-  # end
 end
-
-# target :test do
-#   unreferenced!                     # Skip type checking the `lib` code when types in `test` target is changed
-#   signature "sig/test"              # Put RBS files for tests under `sig/test`
-#   check "test"                      # Type check Ruby scripts under `test`
-#
-#   configure_code_diagnostics(D::Ruby.lenient)      # Weak type checking for test code
-#
-#   # library "pathname"              # Standard libraries
-# end

data/compose.yml

@@ -0,0 +1,33 @@
+services:
+  postgres:
+    image: postgres:16
+    container_name: arfi-postgres
+    environment:
+      POSTGRES_USER: postgres
+      POSTGRES_PASSWORD: postgres
+      POSTGRES_DB: arfi_test
+    ports:
+      - "5432:5432"
+    healthcheck:
+      test: [ "CMD-SHELL", "pg_isready -U postgres -d arfi_test" ]
+      interval: 2s
+      timeout: 3s
+      retries: 30
+    volumes:
+      - arfi_pg_data:/var/lib/postgresql/data
+  mysql:
+    image: mysql:8.0
+    ports:
+      - "3306:3306"
+    environment:
+      MYSQL_ROOT_PASSWORD: password
+      MYSQL_DATABASE: arfi_test
+    healthcheck:
+      test: [ "CMD-SHELL", "mysqladmin ping -h 127.0.0.1 -ppassword --silent" ]
+      interval: 5s
+      timeout: 3s
+      retries: 30
+      start_period: 10s
+
+volumes:
+  arfi_pg_data: