What is being built, what is being weighed, what we have decided not to build, and what has just landed.
This page describes Calíope 1.5. It is the version we are building right now; 1.4 is finished and in App Review, and the store serves 1.3 on the Mac and 1.2 on iPad, so what is marked below as shipped in 1.4 or 1.5 is finished rather than available to download today.
Calíope grows from three things: what arrives through support, the diagnostic reports people choose to share, and daily work against real databases. This page says where each item stands — including the ones we decided not to build, with the reason, because a roadmap that only lists promises is not telling you much. Something missing? Suggest a feature.
Adding an engine to Calíope is implementing a set of protocols, not editing the tools. That work is done and proven: the backup engine, the schema comparator and the error catalogue already run through it, and no engine name is written anywhere outside its own folder — not even in the five translation files. PostgreSQL was the first engine to arrive that way, in 1.3, without a single tool being rewritten. The next one is the same job again.
Microsoft SQL Server as the next engine. The execution plan is already written, against the same seam PostgreSQL came through: adding an engine is implementing a set of protocols, not editing the tools. It sits here rather than in progress on purpose — nothing has been built yet.
An in-memory key-value store, which is a different shape from a table: a key browser with the patterns you actually type, the time to live of each key visible and editable, a viewer that draws every value by its type — string, list, set, sorted set, hash, stream — and a server panel with memory, keyspace and clients. There is no SQL here, so it would arrive as its own tool, the way MongoDB did.
Columnar analytics in a local file, with no server to start. It reads CSV, JSON and Parquet directly with SQL, so a folder of exports becomes something you can query and join without loading it anywhere first. Its place is next to the SQLite editor, not in the connection list: there is no host and no password.
Columnar analytics on a server, over its HTTP interface. What it adds is the scale a row-oriented engine does not reach: aggregations over billions of rows, and materialized views that keep the expensive part precomputed. It reuses the SQL editor and the schema browser, with its own dialect and its own vocabulary for what a table is made of. Between them, the four would take Calíope from three shapes of database to five: relational client-server and relational in a file are here today, documents arrived with MongoDB, and these add key-value and columnar analytics, local and distributed.
Calíope ships in five: English, Spanish, French, German and Italian. Portuguese and Japanese have both come up. Tell us which one you would like next — the app is fully translated rather than machine-localised, so each new language is real work and we would rather do the right one.
The SQLite Editor opens unencrypted databases only. Supporting SQLCipher means shipping and maintaining a fork of SQLite inside a sandboxed App Store app, and the tool is meant for inspecting Core Data stores and embedded application databases, which are not encrypted.
Showing where you are would require a location permission and a matching entry in the App Store privacy label, in exchange for something this tool does not need: what matters on the map is where your data is, not where you are.
It would be tempting to test the main and the alternate host in parallel and pick the fastest. Opening and dropping TCP connections without completing the handshake increases aborted_connects on the server and can end in Host is blocked because of many connection errors. The failover happens on the real attempt instead.
MongoDB is a self-contained tool with its own session and its own set of tabs, not a new engine behind the existing ones. The relational abstraction models foreign keys, binlogs, replication and EXPLAIN; a document database would leave most of that empty, and every tool would have to start asking what kind of session it is looking at.
Calíope lays itself out for the phone and installs from the same App Store page as the Mac and the iPad. It is the whole client, not a viewer: it connects to your servers and runs your queries, with the same tools as the iPad and no device check anywhere in the tree. Your profiles, snippets, history and assistant conversations arrive through your own iCloud, and when that switch is closed it names it instead of showing an empty screen and letting you conclude there is nothing wrong.
Native PostgreSQL connectivity alongside MariaDB, MySQL and Amazon Aurora, with its own provider rather than a MySQL adapter: connection profiles, schema browser, SQL editor with its own snippets and keywords, backups and restores, maintenance, roles and privileges, and server monitoring. Replication is the one area still MySQL-only, and what an engine cannot do is hidden rather than greyed out.
A tool for document databases, with their own vocabulary: browse collections and documents, edit them as JSON, write filters and aggregation pipelines, manage indexes, and check the server, its running operations and its users. Every value shows its type — ObjectId, decimal128, dates, binaries — because in a schemaless database that is half the information.
In progress means we have committed and work is underway.
Under consideration means we are weighing scope, technical fit and demand. No delivery date is implied.
Not planned means we decided against it. The entry stays visible with its reason, so the decision is transparent rather than silent.
Shipped means it is in the app today. The version it landed in is on each entry — and note that the newest version may still be in App Store review.
Looking for the full list of what has shipped? See the changelog →