MongrelDB Fortran Client

Modern Fortran 2018 HTTP client for MongrelDB - embedded+server database with SQL, vector search, full-text search, and AI-native retrieval.
Built on the system curl binary and a bundled JSON parser. No external Fortran libraries required. Builds with fpm.

Package

Surface	Package	Install
Fortran client	MongrelDB-Fortran	build from source with fpm + gfortran

History retention: type-bound history_retention_epochs, earliest_retained_epoch, and set_history_retention_epochs procedures (plus history_retention returning both values).

Requirements

• A Fortran 2018 compiler (gfortran 11+, ifort/flang also work)
• The curl binary on your PATH (the HTTP transport)
• fpm (the Fortran Package Manager, to build)
• A running mongreldb-server daemon

What It Provides

• Typed CRUD over the Kit transaction endpoint: put, upsert (insert-or-update on PK conflict), delete by row id or primary key, with idempotency keys for safe retries.
• Query builder that pushes conditions down to the engine's specialized indexes for sub-millisecond lookups: bitmap equality, learned-range, null checks, and FM-index full-text search. Conditions are AND-ed.
• Idempotent batch transactions - all operations committed atomically, with the engine enforcing unique, foreign key, and check constraints at commit time. Idempotency keys return the original response on duplicate commits, even after a crash.
• Full SQL access through the DataFusion-backed /sql endpoint: recursive CTEs, window functions, CREATE TABLE AS SELECT, materialized views, and multi-statement execution.
• Schema management: typed table creation, full schema catalog, and per-table descriptors.
• Typed error codes: MDB_ERR_AUTH (401/403), MDB_ERR_NOT_FOUND (404), MDB_ERR_CONFLICT (409), MDB_ERR_QUERY (400/5xx), plus MDB_ERR_NETWORK, MDB_ERR_JSON, and MDB_ERR_INVALID_ARG.
• Zero external Fortran dependencies: JSON is handled by a bundled parser (mongreldb_json), and HTTP rides on the system curl binary - no C bindings or libcurl-link complexity.

Examples

Runnable, commented examples live in examples/:

• Quickstart - install, start the daemon, write and run a complete program.
• Transactions - batch commits, idempotency keys, constraint handling.
• Queries - every native condition type and the index it pushes down to.
• SQL - recursive CTEs, window functions, advanced SQL.
• Authentication - bearer token, HTTP Basic, and open modes.
• Errors - error codes, the HTTP-status mapping, and recovery patterns.

Quick Example

program demo
  use mongreldb
  implicit none
  type(mongreldb_client) :: db
  integer :: stat
  character(256) :: errmsg

  ! Connect to a running mongreldb-server daemon.
  call db%connect('http://127.0.0.1:8453', stat, errmsg)

  ! Create a table. Column ids are stable on-wire identifiers.
  call db%create_table('orders', '[' // &
    '{"id":1,"name":"id","ty":"int64","primary_key":true,"nullable":false},' // &
    '{"id":2,"name":"customer","ty":"varchar","primary_key":false,"nullable":false},' // &
    '{"id":3,"name":"amount","ty":"float64","primary_key":false,"nullable":true}]', &
    stat, errmsg)

  ! Insert a row. cells is a flat [colId, value, ...] JSON array.
  call db%put('orders', '[1,1,2,"Alice",3,99.5]', stat, errmsg)

  ! Query by primary key.
  block
    character(:), allocatable :: result
    call db%query('{"table":"orders","conditions":[{"pk":{"value":1}}]}', &
                  result, stat, errmsg)
    print *, trim(result)
  end block
end program

Column JSON can include enum_variants, a static default_value, and a dynamic default_expr ("now" or "uuid"). A static default_value may be a string, integer, boolean, explicit JSON null, or a literal string such as "now"; it is emitted as-is and is not interpreted as an expression. default_expr is the only dynamic-default discriminator and must not be used as an alias for default_value. Native table CHECKs use the optional constraints_json argument:

call db%create_table('orders', columns_json, stat, errmsg, &
  constraints_json='{"checks":[{"id":1,"name":"amount_nonneg","expr":' // &
    '{"Ge":[{"Col":3},{"Lit":{"Float64":0.0}}]}}]}')

A column spec with every default shape:

'{"id":4,"name":"active","ty":"bool","nullable":true,"default_value":true}'
'{"id":5,"name":"notes","ty":"varchar","nullable":true,"default_value":null}'
'{"id":6,"name":"created","ty":"varchar","nullable":true,"default_value":"now"}'
'{"id":7,"name":"updated","ty":"varchar","nullable":true,"default_expr":"now"}'

Build

fpm build --profile release

The library builds two internal modules:

• mongreldb_json - a self-contained JSON parser and serializer.
• mongreldb_http - a curl-backed HTTP transport.

and the public mongreldb module that ties them together.

Test

# Offline unit tests (no server needed) - JSON, encoding, error mapping.
fpm test wire_shape

# Live integration suite (requires mongreldb-server on MONGRELDB_URL).
MONGRELDB_URL=http://127.0.0.1:8453 fpm test live_conformance

The live suite implements the full 16-operation conformance matrix and self-skips when no server is reachable.

API Reference

All methods are bound to the mongreldb_client derived type. Every method takes an intent(out) integer :: stat that is MDB_OK on success and a category code on failure. Most accept an optional errmsg for the detail.

Connection

Method	Description
connect(url, stat, errmsg)	Connect with no auth.
connect_with_token(url, token, stat, errmsg)	Connect with a bearer token.
connect_with_basic_auth(url, user, pass, stat, errmsg)	Connect with HTTP Basic.

Operations

Method	Endpoint	Notes
health()	GET /health	Returns logical; never errors.
tables(stat, errmsg)	GET /tables	Returns list of names.
create_table(name, columns_json, stat, errmsg[, table_id, constraints_json, indexes_json])	POST /kit/create_table	Creates a table with optional constraints and all index definitions.
drop_table(name, stat, errmsg)	DELETE /tables/{name}
count(table, stat, errmsg)	GET /tables/{name}/count	Returns row count.
history_retention(epochs, earliest, stat, errmsg)	GET /history/retention	Returns both retention values.
history_retention_epochs(epochs, stat, errmsg)	GET /history/retention	Returns the configured retention window.
earliest_retained_epoch(earliest, stat, errmsg)	GET /history/retention	Returns the earliest readable epoch floor.
set_history_retention_epochs(value, epochs, earliest, stat, errmsg)	PUT /history/retention	Sets the window; returns post-update values.
put(table, cells_json, stat, errmsg)	POST /kit/txn	Single-op put.
upsert(table, cells_json, update_json, stat, errmsg)	POST /kit/txn	Insert-or-update.
delete(table, row_id, stat, errmsg)	POST /kit/txn	Delete by row id.
delete_by_pk(table, pk_json, stat, errmsg)	POST /kit/txn	Delete by PK value.
transaction(ops_json, results_json, stat, errmsg, idem_key)	POST /kit/txn	Atomic batch.
query(query_json, result_json, stat, errmsg)	POST /kit/query	Native index query.
sql(statement, result_json, stat, errmsg)	POST /sql	Raw SQL (format:json).
schema(stat, errmsg)	GET /kit/schema	Returns catalog JSON.
schema_for(table, stat, errmsg)	GET /kit/schema/{name}	Returns descriptor JSON.

Cells format

Row cells are flat JSON arrays in [col_id, value, col_id, value, ...] order. The column id is the numeric id from create_table, not the name. Example:

[1,42,2,"alice@example.com",3,true]

Error codes

Constant	HTTP	Meaning
MDB_OK (0)	-	Success.
MDB_ERR_AUTH (-1)	401, 403	Bad or missing credentials.
MDB_ERR_NOT_FOUND (-2)	404	Unknown table or row.
MDB_ERR_CONFLICT (-3)	409	Unique constraint violation.
MDB_ERR_QUERY (-4)	400, 5xx	Malformed request or server error.
MDB_ERR_NETWORK (-5)	-	Transport failure.
MDB_ERR_JSON (-6)	-	Response was not valid JSON.
MDB_ERR_INVALID_ARG (-8)	-	Caller-supplied payload was malformed.

Security

• Table names are URL-percent-encoded into path segments.
• Auth credentials are validated to reject CR/LF (header-injection guard).
• The curl transport never follows redirects and never honors proxy env vars.
• Response bodies are capped at 256 MiB (MDB_MAX_RESPONSE_BYTES).

See SECURITY.md for the full model.

Related tools

• Mongrel — Commercial multi-system workbench with native MongrelDB support.
• MongrelDB Viewer — Free, open-source MongrelDB GUI and MCP server.

License

Dual-licensed under MIT OR Apache-2.0. Pick either. Contributions are accepted under the same terms.