Logseq Database DB Version - Play Around

Logseq’s pre-alpha “DB version” mode adds a more structured way to model data—using “classes” (super tags) with configurable properties—so users can turn scattered notes into consistent, queryable records. The biggest practical shift is that properties like title, author, status, dates, and page counts can be defined once on a class and then applied across many pages, reducing the repetitive template-and-property setup that previously required manual steps for each note.

In the demo, a new graph (“test one”) is created, then the familiar journal page remains largely unchanged: indentation, bullet behavior, and scrolling work as before. The meaningful differences show up when creating and managing structured data. A new “new class” option appears, enabling a super-tag-like workflow where the class can be configured with typed properties (for example, text fields, number fields, checkboxes, and date-like fields). The interface also introduces a “configure” popup when hovering over a class-linked item, letting users edit the property schema tied to that class.

The workflow is demonstrated by creating two classes: one for “books” and one for “author.” The author class is set up so book entries can reference authors as linked objects rather than plain text. When a new book page is created from the journal, applying the “books” class prompts property entry through the configured fields. Selecting an author creates or links an author object, and the author’s name becomes accessible as a linked page. The demo highlights both the power and the learning curve: the first property in the class behaves like the object’s name, making some explicit “title/name” properties feel redundant, and the author/tag behavior can be inconsistent depending on how the author is created.

Once properties exist, the DB version’s query tools become central. A query can pull all pages matching a property condition (e.g., all books by a given author) and display results in a table. The demo shows sorting and filtering improvements compared with older “all pages” views—especially the ability to sort by page name and to navigate large sets more effectively. Still, the interface has rough edges: scrolling in result tables can be limited, navigation sometimes forces exiting and restarting, and the table layout can feel cluttered with system fields like created/updated/tags.

The demo also touches on schema evolution: renaming or deleting properties in the class template updates existing pages in a “safe proof” way. Deleting a property removes it from the template, but existing values may persist in some views, while new pages follow the updated schema. Finally, the demo shows that web-based features like PDF upload and text highlighting/referencing appear to work in this mode, though some highlight types (like an “area highlight”) seem missing or not visible.

Overall, the pre-alpha DB version aims to make Logseq’s data modeling more consistent and scalable: define classes with typed properties, apply them to pages, link objects like authors, and query across the resulting structure—while the UI and edge cases remain under active refinement.