v1.5.2: MariaDB in the database grid, engine cards checked against 1.0.0

MariaDB now has its own place in the database grid on the home page and on /databases. It connects through the MySQL driver, so it is shown beside the forty-one engines without being counted among them, and its card links to the MySQL provider documentation.

The engine cards were checked again against LibreDB Studio 1.0.0. Agent mode runs statements on PostgreSQL, MySQL and MariaDB, SQLite, DuckDB, SQL Server, MongoDB and Kubernetes, and the cards, the FAQ and the feature limits now say so. Elasticsearch, OpenSearch, DuckDB, SQL Server, Snowflake, SNMP and MySQL state the limits their provider documentation states, and the FAQ counts sixty-eight named engines, as the grid does.

v1.5.1 — the waiting list asks for the license domain

The LibreDB Platform waiting list now asks for the two things a license is issued for: the organization, as it should appear on the license, and the one domain Platform will run on.

The domain field accepts a domain name only, such as libredb-platform.example.com, and its hint points to the installer setting that takes the same value. Requests now arrive with everything needed to issue a key, so nobody has to send a second email to ask for the domain.

v1.5.0 — a page for LibreDB Platform, with a one-minute tour

LibreDB Platform gets its own door on the site. "Platform" now sits in the header and under the install line on the homepage, and both lead to a rewritten /platform page.

The page says what Platform does in plain terms: single sign-on through Keycloak, database credentials sealed so users never see them, access granted and revoked person by person, six roles, and an audit log of every query. The limits are stated next to them, as everywhere else on the site. Platform is no longer described as a beta. The license is free for 1 admin and 3 users for 12 months, and more seats are paid.

A one-minute tour shows an admin granting a database to an analyst and taking it away again. It plays from YouTube, but only once the visitor presses play. Until then the page shows our own poster image and makes no request to Google.

At the bottom of the page is a waiting list. It opens the visitor's own email app with a message to info@libredb.org, and nothing is sent from the site itself. The privacy policy now covers both the YouTube player and these emails.

This release also brings the site up to LibreDB Studio 1.0.0, including the forty-one engines that ship with it.

v1.4.0: studio 0.18.2 and a full fact check

The site now describes studio 0.18.2: 27 engines plus the embedded LibreDB store (nine new engine cards), 46 distribution channels with 39 live, and every page checked against the 0.18.2 release, with the claims that had drifted corrected.

v1.3.9: one feed per engine

Each engine archive on the blog now has its own RSS feed, for example /blog/engine/mysql/rss.xml, and the feeds validate as RSS 2.0 (the author's name moved out of , which RSS defines as an email address). The site-wide /rss.xml keeps the same URL.

v1.3.8: Elasticsearch engine description

The engines page now says the Elasticsearch connection uses Elasticsearch SQL, matching the provider. Nothing else on the site changes.

v1.3.7: GreatSQL application files

Two PDF files for GreatSQL's product compatibility review now sit under an unlisted path in public/cert/. Their mail server dropped the attachments twice and the reviewer asked for a download link. No page links to them and nothing else on the site changes.

v1.3.6: studio 0.17.0, with Prometheus and Kafka

The site now describes LibreDB Studio 0.17.0, which shipped on 26 September. The version chip, the badge and the schema.org version on every page had still been saying 0.16.2.

0.17.0 added two engines, Prometheus and Apache Kafka, and both now have a card on the homepage and on /databases, under two new categories, time-series and streaming. Each one states its limit like every other engine: both are read-only. Prometheus is queried through PromQL over the read APIs, and Kafka is browsed by offset or by time without producing records.

The engine count moves from seventeen to nineteen everywhere it is written. The FAQ now explains why the product listings say eighteen: the site counts LibreDB itself, the embedded store the playground runs on, and the listings count only the engines you point Studio at.

Trino is now called by its own name. It is not an Apache Software Foundation project, and the product repository corrected this on 27 September.

v1.3.5: twelve database projects list LibreDB Studio

Adds a news post: twelve database projects now list LibreDB Studio in their official documentation or ecosystem pages.

v1.3.4: a corrected count in the VictoriaMetrics post

Corrects a count in the VictoriaMetrics post: the MetricsQL code block lists six expressions, and the text now says six where it said seven.