feat(dialect): add MariaDB + Snowflake to playground, fix WASM dialectMap, unify config validation (#432)

* fix(playground): move useState/useCallback above early returns to fix React error #310

useState(copied) and useCallback(handleCopy) were declared after the
if (loading) and if (error) early returns, violating the Rules of Hooks.

When the WASM binary was a 404, the component always hit the error branch
so both renders exited with the same hook count. After committing the WASM
binary (#423), the component now successfully transitions loading→ready,
causing React to see 11 hooks instead of 9 on the second render → #310.

Fix: move both hooks above all conditional returns.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* feat(dialect): add DialectMariaDB constant to keyword dialect system

* fix(dialect): return MYSQL_SPECIFIC for DialectMariaDB and add to validDialects test

* fix(dialect): wire DialectMariaDB into keywords.New() to load MYSQL_SPECIFIC keywords

* test(dialect): add TestDialectMariaDB_InheritsMySQL to guard MySQL keyword inheritance

* feat(dialect): add MARIADB_SPECIFIC keyword list extending MySQL dialect

* feat(dialect): add MariaDB auto-detection hints (SEQUENCE, VERSIONING, CONNECT BY)

* fix(dialect): remove over-broad START WITH hint and complete DetectDialect doc comment

* fix(dialect): restore MariaDB CONNECT BY hint and add accumulation test

* feat(ast): add CreateSequenceStatement, DropSequenceStatement, AlterSequenceStatement nodes

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(ast): nil guard in sequence ToSQL, split Restart field, add sequence tests

- Guard s.Name nil dereference in CreateSequenceStatement.ToSQL,
  DropSequenceStatement.ToSQL, and AlterSequenceStatement.ToSQL
- Split SequenceOptions.Restart (*LiteralValue) into two fields:
  Restart bool (bare RESTART) and RestartWith *LiteralValue (RESTART WITH n)
- Update writeSequenceOptions to emit bare RESTART or RESTART WITH n accordingly
- Add ast_sequence_test.go with full ToSQL table-driven tests and pool round-trip test

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* feat(ast): add ForSystemTimeClause, PeriodDefinition, temporal table fields

- Add SystemTimeClauseType enum with AS OF, BETWEEN, FROM/TO, ALL variants
- Add ForSystemTimeClause struct for MariaDB FOR SYSTEM_TIME temporal queries
- Add PeriodDefinition struct for PERIOD FOR clauses in CREATE TABLE
- Extend TableReference with ForSystemTime field (MariaDB 10.3.4+)
- Extend CreateTableStatement with WithSystemVersioning and PeriodDefinitions fields
- Add ForSystemTimeClause.ToSQL() and tableRefSQL integration in sql.go

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(ast): add SQL() methods to temporal/CONNECT BY types and fix SelectStatement field order

Add SQL() methods to ForSystemTimeClause, ConnectByClause, and PeriodDefinition
(all implement expressionNode()) so they satisfy the Expression interface fully
without silently degrading via the exprSQL() fallback. Move StartWith and
ConnectBy fields in SelectStatement to directly follow Having, matching logical
SQL clause ordering.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* feat(parser): implement CREATE/DROP/ALTER SEQUENCE parsing for MariaDB dialect

Add parseCreateSequenceStatement, parseDropSequenceStatement, and
parseAlterSequenceStatement to mariadb.go with full option parsing
(START WITH, INCREMENT BY, MINVALUE, MAXVALUE, CYCLE, CACHE, RESTART WITH).
Wire dispatch into parseStatement() for DROP/ALTER and into parseCreateStatement()
for CREATE. Gate all paths behind isMariaDB() so MySQL and other dialects
are unaffected. Add six passing parser tests in mariadb_test.go.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* feat(parser): implement temporal table parsing (FOR SYSTEM_TIME, WITH SYSTEM VERSIONING, PERIOD FOR)

Add parseForSystemTimeClause and parseTemporalPointExpression to mariadb.go
supporting AS OF, BETWEEN, FROM/TO, and ALL variants. Hook into
parseFromTableReference in select_subquery.go (after alias, before SQL Server
hints) with a peek-ahead guard so FOR is only consumed when followed by
SYSTEM_TIME. Add WITH SYSTEM VERSIONING parsing to parseCreateTable (after
closing paren, before PARTITION BY) and PERIOD FOR column parsing to the
column loop in ddl.go. Add four passing tests for temporal queries and
system versioning in mariadb_test.go.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* feat(parser): implement CONNECT BY hierarchical query parsing for MariaDB dialect

Add parseConnectByCondition to mariadb.go which handles the PRIOR prefix
operator by wrapping the referenced column in a FunctionCall node and
building a BinaryExpression for the full PRIOR col = parent_col pattern.
Wire START WITH and CONNECT BY [NOCYCLE] parsing into parseSelectStatement
in select.go after the HAVING clause. Guard CONNECT and START from being
consumed as table aliases in parseFromTableReference via a peek-ahead check
in select_subquery.go. Add three passing tests covering basic, NOCYCLE, and
no-START-WITH variants in mariadb_test.go.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(parser): add bare RESTART test and nil guard for CONNECT BY condition

* test(parser): add MariaDB SQL test data files and file-based integration test

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* docs: add MariaDB dialect to SQL_COMPATIBILITY.md and CHANGELOG.md

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(mariadb): address code review issues across AST, keywords, and parser

AST layer:
- Add `Prior` UnaryOperator constant for CONNECT BY PRIOR expressions
- Add `NoCache bool` to SequenceOptions (NOCACHE was previously a silent no-op)
- Add `Pos models.Location` to all 6 new MariaDB AST nodes (CreateSequence,
  DropSequence, AlterSequence, ForSystemTimeClause, ConnectByClause, PeriodDefinition)
- Add `NewDropSequenceStatement()` / `ReleaseDropSequenceStatement()` pool funcs
  to match Create/Alter sequence pooling consistency
- Emit NOCACHE in writeSequenceOptions when NoCache=true

Keywords:
- Remove duplicate MINVALUE, MAXVALUE, INCREMENT, RESTART, NOCACHE entries
  from MARIADB_SPECIFIC (already covered by base/Oracle keyword lists)

Parser:
- Fix PRIOR operator: switch from FunctionCall to UnaryExpression{Operator: Prior}
- Fix PRIOR on RHS: CONNECT BY col = PRIOR col now parsed correctly
- Fix parseJoinedTableRef: add isMariaDBClauseKeyword guard to prevent CONNECT/START
  from being consumed as table aliases (same guard already in parseFromTableReference)
- Fix parseJoinedTableRef: add FOR SYSTEM_TIME temporal clause support on JOIN refs
- Fix DROP SEQUENCE: support IF NOT EXISTS in addition to IF EXISTS
- Fix DROP SEQUENCE: use NewDropSequenceStatement() for pool consistency
- Fix parseSequenceOptions: set opts.NoCache=true for NOCACHE keyword
- Add comment in parseTemporalPointExpression explaining quote-stripping behaviour

Tests:
- Add TestMariaDB_ConnectBy_PriorOnRight
- Add TestMariaDB_DropSequence_IfNotExists
- Add TestMariaDB_Sequence_NoCache
- Expand testdata SQL files with NO MINVALUE/MAXVALUE forms, PRIOR-on-right cases,
  IF NOT EXISTS on DROP, and multi-table temporal JOIN query

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(mariadb): address second code review pass — Pos, NO CACHE, CONNECT BY, dedup

Parser dispatch (parser.go, ddl.go, select.go):
- Populate Pos on CreateSequenceStatement (at SEQUENCE token in ddl.go)
- Populate Pos on DropSequenceStatement (at DROP token in parser.go)
- Populate Pos on AlterSequenceStatement (at ALTER token in parser.go)
- Populate Pos on ConnectByClause (at CONNECT token in select.go)
- Populate Pos on PeriodDefinition (at PERIOD token in ddl.go)

mariadb.go:
- Fix NO CACHE (two-token) to also set opts.NoCache=true, matching NOCACHE
- Fix parseConnectByCondition to handle complex AND/OR chains:
  CONNECT BY PRIOR id = parent_id AND active = 1 now fully parsed
- Extract isMariaDBClauseStart() method (was duplicated closure in two functions)
- Populate Pos on ForSystemTimeClause (at SYSTEM_TIME token)
- Add comment clarifying IF NOT EXISTS is a non-standard permissive extension

select_subquery.go:
- Remove both isMariaDBClauseKeyword closures, replace with p.isMariaDBClauseStart()

ast.go:
- Update DropSequenceStatement doc to show [IF EXISTS | IF NOT EXISTS]

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(mariadb): correct START WITH/CONNECT BY SQL order and implement PeriodDefinition.SQL()

- Move START WITH / CONNECT BY emission to after HAVING, before ORDER BY
- Implement PeriodDefinition.SQL() — was silently returning empty string
- Add round-trip tests for both fixes

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(mariadb): code review fixes — pool comments, expressionNode docs, PERIOD FOR SYSTEM_TIME test

- Add '// zero all fields' comment to ReleaseCreateSequenceStatement and ReleaseAlterSequenceStatement
- Add design-choice comments on expressionNode() for ForSystemTimeClause, ConnectByClause, PeriodDefinition
- Add TestMariaDB_CreateTable_PeriodForSystemTime to cover PERIOD FOR SYSTEM_TIME parsing
- Fix parsePeriodDefinition to use parseColumnName so SYSTEM_TIME (reserved keyword) is accepted as period name
- Add GENERATED ALWAYS AS ROW START/END column constraint parsing in parseColumnConstraint (MariaDB system-versioned columns)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(dialect): reduce MariaDB CONNECT BY hint weight to 2 so Oracle wins pure CONNECT BY queries

Oracle's CONNECT BY weight is 3; MariaDB's was also 3, causing a tie broken
by non-deterministic map iteration. Reducing MariaDB to weight 2 ensures
Oracle wins when CONNECT BY is the only hint. MariaDB is still correctly
detected when its unique high-weight hints (NEXTVAL, FOR SYSTEM_TIME, etc.)
are present.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(mariadb): address code review — CycleOption enum, CACHE validation, structured errors

- Replace Cycle bool + NoCycle bool with CycleOption enum to prevent
  invalid state where both could be true simultaneously
- Add validation in parseSequenceOptions for contradictory CACHE/NOCACHE
- Replace fmt.Errorf with p.expectedError() for consistent error style

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* bench(mariadb): add MariaDB-specific parsing benchmarks

17 benchmarks across 4 groups:
- BenchmarkMariaDB_Sequence: CREATE/ALTER/DROP SEQUENCE (5 cases)
- BenchmarkMariaDB_ForSystemTime: temporal table queries (5 cases)
- BenchmarkMariaDB_ConnectBy: hierarchical queries with PRIOR (4 cases)
- BenchmarkMariaDB_Mixed: combined features — CTE+temporal, CTE+CONNECT BY,
  CREATE TABLE WITH SYSTEM VERSIONING (3 cases)

Baseline (M2): 398–3001 ns/op, 984–6763 B/op

* feat(playground): add Snowflake and MariaDB to dialect dropdown

Closes #428 — Snowflake was missing despite full keyword support.
Closes #426 (partial) — MariaDB playground integration complete.
Playground now exposes all 8 named dialects + Generic (9 total).

* fix(wasm): add mariadb to dialectMap so playground selection takes effect

Without this, selecting MariaDB in the dropdown silently fell back to
generic dialect parsing. The UI entry was a no-op at the parser layer.

* fix(wasm): add clickhouse to dialectMap (pre-existing silent fallback bug)

* fix(config): add snowflake to validDialects and update help text

snowflake was accepted by --dialect flag but rejected when set in
YAML config. Both layers are now consistent.

Part of #428.

* fix(config): add snowflake+mariadb to internal ValidDialects, tests, and doc comments

Add "snowflake" and "mariadb" to the ValidDialects slice and inline
validDialects list in cmd/gosqlx/internal/config, update corresponding
tests, and sync doc comments in pkg/config to reflect all supported
dialects including the two new ones.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(config): update remaining dialect doc strings and LSP runtime description

* docs: update dialect counts 7→8 and add MariaDB to all doc lists

- CLAUDE.md: add Snowflake + MariaDB (was missing both)
- README.md: 7→8 dialects in stats bar, add MariaDB to feature lists
- SQL_COMPATIBILITY.md: 6→8 Supported Dialects (add ClickHouse + MariaDB)
- website/src/lib/constants.ts: add MariaDB to feature description

Resolves #426 (docs integration)
Resolves #428 (count sync)

* docs: update dialect counts in publishing drafts (7→8, add MariaDB)

* chore: remove plan files from PR (internal docs)

---------

Co-authored-by: Ajit Pratap Singh <ajitpratapsingh@Ajits-Mac-mini-2655.local>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: 3 people

CLAUDE.md

@@ -12,7 +12,7 @@ GoSQLX is a **production-ready**, **race-free**, high-performance SQL parsing SD
 - Thread-safe with zero race conditions (20,000+ concurrent operations tested)
 - 1.38M+ ops/sec sustained, 1.5M peak with memory-efficient object pooling
 - ~80-85% SQL-99 compliance (window functions, CTEs, set operations, MERGE, etc.)
-- Multi-dialect support: PostgreSQL, MySQL, SQL Server, Oracle, SQLite, ClickHouse
+- Multi-dialect support: PostgreSQL, MySQL, MariaDB, SQL Server, Oracle, SQLite, Snowflake, ClickHouse (8 dialects)
 
 ## Architecture
 

README.md

@@ -28,7 +28,7 @@
 
 <br/>
 
-| **1.38M+ ops/sec** | **<1μs latency** | **85% SQL-99** | **7 dialects** | **0 race conditions** |
+| **1.38M+ ops/sec** | **<1μs latency** | **85% SQL-99** | **8 dialects** | **0 race conditions** |
 |:---:|:---:|:---:|:---:|:---:|
 
 </div>
@@ -48,7 +48,7 @@ ast, _ := gosqlx.Parse("SELECT u.name, COUNT(*) FROM users u JOIN orders o ON u.
 
 - **Not an ORM** - a parser. You get the AST, you decide what to do with it.
 - **Not slow** - zero-copy tokenization, sync.Pool recycling, no allocations on hot paths.
-- **Not limited** - PostgreSQL, MySQL, SQL Server, Oracle, SQLite, Snowflake, ClickHouse. CTEs, window functions, MERGE, set operations.
+- **Not limited** - PostgreSQL, MySQL, MariaDB, SQL Server, Oracle, SQLite, Snowflake, ClickHouse. CTEs, window functions, MERGE, set operations.
 - **Not just a library** - CLI, VS Code extension, GitHub Action, MCP server, WASM playground, Python bindings.
 
 <br/>
@@ -140,7 +140,7 @@ claude mcp add --transport http gosqlx \
 <td align="center" width="33%"><h3>🔧 Tooling</h3>AST-based formatter<br/>Query transforms API<br/>VS Code extension<br/>GitHub Action</td>
 </tr>
 <tr>
-<td align="center"><h3>🌐 Multi-Dialect</h3>PostgreSQL · MySQL<br/>SQL Server · Oracle<br/>SQLite · Snowflake</td>
+<td align="center"><h3>🌐 Multi-Dialect</h3>PostgreSQL · MySQL · MariaDB<br/>SQL Server · Oracle<br/>SQLite · Snowflake · ClickHouse</td>
 <td align="center"><h3>🤖 AI-Ready</h3>MCP server (7 tools)<br/>Public remote endpoint<br/>Streamable HTTP</td>
 <td align="center"><h3>🧪 Battle-Tested</h3>20K+ concurrent ops<br/>Zero race conditions<br/>~85% SQL-99 compliance</td>
 </tr>

cmd/gosqlx/cmd/validate.go

@@ -89,7 +89,7 @@ func validateRun(cmd *cobra.Command, args []string) error {
 
 	// Reject unknown dialect names early before any parsing.
 	if validateDialect != "" && !keywords.IsValidDialect(validateDialect) {
-		return fmt.Errorf("unknown SQL dialect %q; valid dialects: postgresql, mysql, sqlserver, oracle, sqlite, snowflake, bigquery, redshift", validateDialect)
+		return fmt.Errorf("unknown SQL dialect %q; valid dialects: postgresql, mysql, mariadb, sqlserver, oracle, sqlite, snowflake, bigquery, redshift", validateDialect)
 	}
 
 	// Handle stdin input
@@ -334,7 +334,7 @@ func init() {
 	validateCmd.Flags().BoolVarP(&validateQuiet, "quiet", "q", false, "quiet mode (exit code only)")
 	validateCmd.Flags().BoolVar(&validateQuiet, "check", false, "check mode (alias for --quiet): exit code only, no output")
 	validateCmd.Flags().BoolVarP(&validateStats, "stats", "s", false, "show performance statistics")
-	validateCmd.Flags().StringVar(&validateDialect, "dialect", "", "SQL dialect: postgresql, mysql, sqlserver, oracle, sqlite (config: validate.dialect)")
+	validateCmd.Flags().StringVar(&validateDialect, "dialect", "", "SQL dialect: postgresql, mysql, mariadb, snowflake, sqlserver, oracle, sqlite (config: validate.dialect)")
 	validateCmd.Flags().BoolVar(&validateStrict, "strict", false, "enable strict validation mode (config: validate.strict_mode)")
 	validateCmd.Flags().StringVar(&validateOutputFormat, "output-format", "text", "output format: text, json, sarif")
 	validateCmd.Flags().StringVar(&validateOutputFile, "output-file", "", "output file path (default: stdout)")

cmd/gosqlx/internal/config/config.go

@@ -84,7 +84,7 @@ type FormatConfig struct {
 // Controls validation behavior including dialect selection and security limits.
 //
 // Fields:
-//   - Dialect: SQL dialect for validation (postgresql, mysql, sqlserver, oracle, sqlite, generic)
+//   - Dialect: SQL dialect for validation (postgresql, mysql, mariadb, snowflake, sqlserver, oracle, sqlite, generic)
 //   - StrictMode: Enable strict validation rules (default: false)
 //   - Recursive: Recursively process directories (default: false)
 //   - Pattern: File pattern for recursive processing (default: "*.sql")
@@ -345,7 +345,7 @@ func (c *Config) Validate() error {
 
 	// Validate dialect (empty string means permissive/no dialect gates)
 	if c.Validation.Dialect != "" {
-		validDialects := []string{"postgresql", "mysql", "sqlserver", "oracle", "sqlite", "generic"}
+		validDialects := []string{"postgresql", "mysql", "mariadb", "snowflake", "sqlserver", "oracle", "sqlite", "generic"}
 		dialectValid := false
 		for _, d := range validDialects {
 			if c.Validation.Dialect == d {

cmd/gosqlx/internal/config/doc.go

@@ -152,7 +152,7 @@
 //	    max_file_size: 10485760    # Maximum file size in bytes
 //
 // Fields:
-//   - Dialect: SQL dialect (postgresql, mysql, sqlserver, oracle, sqlite, generic)
+//   - Dialect: SQL dialect (postgresql, mysql, mariadb, snowflake, sqlserver, oracle, sqlite, generic)
 //   - StrictMode: Enable strict validation rules (default: false)
 //   - Recursive: Recursively process directories (default: false)
 //   - Pattern: File pattern for recursive processing (default: "*.sql")
@@ -195,7 +195,7 @@
 //   - MaxLineLength: 0-500 characters
 //
 // Validation validation:
-//   - Dialect: Must be one of: postgresql, mysql, sqlserver, oracle, sqlite, generic
+//   - Dialect: Must be one of: postgresql, mysql, mariadb, snowflake, sqlserver, oracle, sqlite, generic
 //
 // Output validation:
 //   - Format: Must be one of: json, yaml, table, tree, auto

cmd/gosqlx/internal/config/schema.go

@@ -23,6 +23,8 @@ import (
 var ValidDialects = []string{
 	"postgresql",
 	"mysql",
+	"mariadb",
+	"snowflake",
 	"sqlserver",
 	"oracle",
 	"sqlite",

cmd/gosqlx/internal/config/schema_test.go

@@ -26,6 +26,8 @@ func TestValidateDialect(t *testing.T) {
 	}{
 		{"valid postgresql", "postgresql", false},
 		{"valid mysql", "mysql", false},
+		{"valid mariadb", "mariadb", false},
+		{"valid snowflake", "snowflake", false},
 		{"valid sqlserver", "sqlserver", false},
 		{"valid oracle", "oracle", false},
 		{"valid sqlite", "sqlite", false},
@@ -225,7 +227,7 @@ func TestGetSchema(t *testing.T) {
 }
 
 func TestValidDialects(t *testing.T) {
-	expectedDialects := []string{"postgresql", "mysql", "sqlserver", "oracle", "sqlite", "generic"}
+	expectedDialects := []string{"postgresql", "mysql", "mariadb", "snowflake", "sqlserver", "oracle", "sqlite", "generic"}
 
 	if len(ValidDialects) != len(expectedDialects) {
 		t.Errorf("expected %d dialects, got %d", len(expectedDialects), len(ValidDialects))

docs/SQL_COMPATIBILITY.md

@@ -862,7 +862,7 @@ gosqlx format --dialect mysql query.sql
 1. **ParseWithDialect()** - Parse SQL with dialect-specific syntax
 2. **ValidateWithDialect()** - Validate with dialect awareness
 3. **--dialect CLI flag** - Specify dialect for CLI commands
-4. **6 Supported Dialects** - PostgreSQL, MySQL, SQL Server, Oracle, SQLite, Snowflake
+4. **8 Supported Dialects** - PostgreSQL, MySQL, MariaDB, SQL Server, Oracle, SQLite, Snowflake, ClickHouse
 
 ### MySQL Syntax (11 Features)
 1. **SHOW statements** - SHOW TABLES, DATABASES, CREATE TABLE

pkg/config/config.go

@@ -77,12 +77,12 @@ type FormatConfig struct {
 // ValidationConfig holds SQL validation options for the parser and validator.
 //
 // The Dialect field determines which SQL keywords and syntax are recognized.
-// Supported values: "postgresql", "mysql", "sqlserver", "oracle", "sqlite".
+// Supported values: "postgresql", "mysql", "mariadb", "snowflake", "sqlserver", "oracle", "sqlite".
 //
 // The Pattern field is used for recursive file validation and supports standard
 // glob patterns like "*.sql", "queries/**/*.sql", etc.
 type ValidationConfig struct {
-	Dialect    string         `yaml:"dialect" json:"dialect"`        // SQL dialect: postgresql, mysql, sqlserver, oracle, sqlite (default: "postgresql")
+	Dialect    string         `yaml:"dialect" json:"dialect"`        // SQL dialect: postgresql, mysql, mariadb, snowflake, sqlserver, oracle, sqlite (default: "postgresql")
 	StrictMode *bool          `yaml:"strict_mode" json:"strictMode"` // Enable strict validation mode (default: false)
 	Recursive  *bool          `yaml:"recursive" json:"recursive"`    // Recursively validate files in directories (default: false)
 	Pattern    string         `yaml:"pattern" json:"pattern"`        // File pattern for recursive validation (default: "*.sql")
@@ -215,12 +215,14 @@ func (c *Config) Validate() error {
 	validDialects := map[string]bool{
 		"postgresql": true,
 		"mysql":      true,
+		"mariadb":    true,
+		"snowflake":  true,
 		"sqlserver":  true,
 		"oracle":     true,
 		"sqlite":     true,
 	}
 	if c.Validation.Dialect != "" && !validDialects[c.Validation.Dialect] {
-		return fmt.Errorf("validation.dialect must be one of: postgresql, mysql, sqlserver, oracle, sqlite; got %q", c.Validation.Dialect)
+		return fmt.Errorf("validation.dialect must be one of: postgresql, mysql, mariadb, snowflake, sqlserver, oracle, sqlite; got %q", c.Validation.Dialect)
 	}
 	if c.Validation.Security.MaxFileSize < 0 {
 		return fmt.Errorf("validation.security.max_file_size must be non-negative, got %d", c.Validation.Security.MaxFileSize)

pkg/config/config_test.go

@@ -231,7 +231,7 @@ func TestConfigMerge(t *testing.T) {
 }
 
 func TestConfigDialects(t *testing.T) {
-	validDialects := []string{"postgresql", "mysql", "sqlserver", "oracle", "sqlite"}
+	validDialects := []string{"postgresql", "mysql", "mariadb", "snowflake", "sqlserver", "oracle", "sqlite"}
 
 	for _, dialect := range validDialects {
 		t.Run(dialect, func(t *testing.T) {