What I am missing from Tana

Hi! The title speaks for itself:

  1. Everything is a node
  2. Supertags and their inheritance
  3. Filters / “Smart folders”
  4. Semantic page relations / recursive searching

Inspired by article Tana: The Ultimate Note-Taking App — A Review of its Key Features and Innovations
- not affiliated with the author or Tana.
I would be interested in developers’ vision, which of the features and priority are on their roadmap. All of them would make Logseq an even greater product in my opinion. Below short description:

1. Everything is a node
This will reduce mental friction of categorizing new notes into either page or block structure. Just dump a block in the journal and later decide if this block should get a name, “upgrading” it to a page. I use separate pages mainly to get global names and to store short metadata, this would mean less markdown files to deal with.

2. Supertags and their inheritance
It is a bit like Logseq templates, in a more dynamic way. Being able to define a Book template page with author property, and let Logseq autofill author and enforce an [[Author]] value, for every new created Book note. Changes in the base page enforce structure changes in all subpages. Science-fiction book can be subkind of Book, inherit all its properties and add new ones, which let’s us add more structure to data. (Apparently some work has been done, but no news since then)

3. Filters / “Smart folders”
Could be called “Smart pages” in Logseq. A good example is to create a smart page for URLs, like https://discuss.logseq.com and give it a name Logseq forum discussions. Logseq auto-searches text content containing URL in the graph and makes those blocks entries of the page. Logseq forum discussions behaves like a normal page.

4. Semantic page relations / recursive searching
See https://twitter.com/houshuang/status/1603372657549811712 for how this looks like in Tana. Basic idea is to define page relations by using properties and being able to search across multiple pages by looking at their relation. Going with the book example, search for Books also finds Science-Fiction books (IS-A).

+1 upvote on smart tags and inheriting predefined property fields.

2 Likes

Check out remnote. Everything is essentially a rem, or a node, and they can be turned into pages or folders. The huge win for me is you can organize everything from the top down. Everything can be one big giant list, and you just continue drilling down further and further.

I have not come across anything in remnote that mimics supertags, but I am still fairly new to the application and may not be aware if this is a possibility. I do know they are adding tables and databases to remnote which are currently in alpha testing, and allow for rem’s to contain properties.

Each page also has a filter in the top right corner where you can apply filters on the page. It also supports what are called “portals” which are similar to a query in Logseq or Tana, and allows you to embed these portals inside a rem.

Lastly, it also supports local storage, so your data is stored on your local system. It’s stored in a database file, and the application gives you the ability to export out to markdown or other formats.

I don’t think Logseq will change the storing model based on pages but the same UX of Tana could be implemented in two steps:

  1. Given a page reference in a block, provide a way to quickly display the page content as children blocks.
  2. Provide a hierarchical overview of pages without the need to manually compile a Contents page.
1 Like

Not sure, what you have in mind. Maybe a more modal peek view when mouse-hovering page links to edit page content in-place? I also could imagine a temporary embedded view below block with focused page link similar to embedded queries, to make content accessible for keyboard.

This seems to be an orthogonal feature - which I would second as well for usage with current version.


Afterthought
First point “1. Everything is a node” has become a bit more neglectable for me after having heard of planned database version:

  • Amount of pages and size of their content does not matter, since everything is stored in database
  • Choice of characters for page title is more permissive than filesystem (inline code via backtick ` would be great to have in title)

Points 2-4 still seem very useful to have.

Blocks that have been referenced elsewhere have a button on the right that indicates the number of references and by clicking it you can show the references below. I mean the same thing but displaying the blocks of a referenced page.

is a pretty nice feature in the latest Obsidian, looks like a step towards Tana features.

Am I correct that this is just for page properties? I think the major advantage of Logseq properties (and Tana’s fields) is that they are for blocks in an outliner UI.

Yes, it is just for page properties. If one clicks on the button to set a property, it is set at the top of the page, yaml-like.

It is more basic than what Logseq has. It has no block properties, and I don’t know how sophisticated the queries are.