Repository navigation
Replies: 3 comments 1 reply
|
One possible approach: authorship inherits down the tree. I sketched this earlier in #645 (#645 (comment)). Rule. A block's author is the nearest The same page as above carries the same information with three On write, roughly:
Open questions I haven't worked out:
This is a direction, not a finished design. Happy to hear other ways to do it. |
|
Why not just use git blame and then just wrap git blame. |
|
I'm writing here in relation to issue #846, where I propose making author optional. Regarding your questions, though, I don't think this is something specific to Tine, but rather something that depends on each person's own model. To illustrate this, let me translate it into a different use case. Context: You work on two projects (A, B) that have related tasks (a task in A affects a task in B, and vice versa), and several people are involved (Carlo, Anna). Consider this block: The hierarchy establishes dependencies between tasks (the blocks in an outliner), but beyond that, it's all about each person's own model. The first TODO establishes that Anna has to develop feature 1 within project A. The same applies to the last TODO, feature 7 (through inheritance). Feature 3 belongs to project B, but Anna also has to work on it. (Here, I could adopt a different idea than feature 7: if there's no person tag, then nobody is assigned. But that's not how my model works. Inheritance first) Feature 4 is for Carlo within project A. Feature 5 is for Anna, indicating that Carlo isn't the one doing it, but there is a task dependency between Carlo's work and Anna's. Finally, feature 6 is done by Carlo within project B. It might seem a bit complicated, but I've been doing it this way for years, so you get used to it. Also, if you use MS Project, this is essentially how you model tasks within projects. Now, to answer your questions:
In my model, yes, and that's what I do. If I want something more robust and foolproof, I would add the tags #Carlo, #Anna, #projectA, and #projectB to every block. That way, I'd never have any problems when rearranging tasks (imagine tasks belonging to a real project that lasts several years; things can get complicated). But it's up to each person to decide how they want to manage this. Another thing: sometimes I capture tasks in journals and use block references that survive when blocks are moved around. That's another approach.
Use block references. It also depends on what you want: when you copy and paste, do you want to change the author or not? Sometimes I copy and paste because it's the same task in two different projects. For example:
I copy it to:
Now Anna has to develop feature 1 in both project A and project B. Again, it depends on each person's approach.
Same idea as with copy and paste. You decide: do you want the author to become yourself or remain someone else? You reflect that in your model. For example, I sometimes do this:
What do I gain from this? When I visit Carlo's page, I can see the blocks assigned to him. And since it's a hierarchy, under assigned he can see all the blocks that have been assigned to him.
And here's the holy grail! I use Logseq with advanced queries to support my model. Tine has improved things in this area and supports hierarchical queries (ancestor and descendant blocks), so this should work for you too. Everything I've described is possible in Logseq. The great thing about Logseq is that it doesn't force much on you beyond the bare minimum, allowing you to build your own model on top of it. I hope this helps. I ended up here because Claude directed me here after I expressed my wish to make |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Authorship is useful. With agents writing into my graph, I want to see at a glance which text is mine and which isn't. But the way it's stored now doesn't scale with how much an author writes.
tine addputsauthor::on every block it creates. When an agent writes 100 blocks, the file gets 100 extra property lines that all say the same thing. That causes three problems:author::. Reading the raw Markdown or a git diff gets harder.Example: my own page, where Claude added a plan and Codex added one line later.
Nine content lines, eight
author::lines. The only fact they carry is "Claude wrote Day 1 and Day 2, Codex wrote one line".How do others see this? Is per-block authorship what you want, or should it be stored more compactly? I'll post one possible approach in a comment below.
All reactions