Logseq frustrates me, and while it can be fixed around the edges (I even got a PR feature in) its basic model is just not that great, and I don't think it would be easy for me to change in significant ways.
See NextPKM and NextPKM/Toolkit and NextPKM/Planning OK this is the third or fourth time I've writen down the same ideas.
Instant-publish, no batch processing with goddinpotty, that all is incorporated
Not even sure that is a good idea, but operationally, really would be a good thing.
No separate graphs, although you can do the equivalent with tagging or whatever.
eg, a window which filters on a tag and all new blocks get the tag: that's a separate graph right there!
Tags doing a lot of work
Eg, set a different look for different tones. Sounds kind of hokey but I really need something like this, just to keep track of who I am talking to.
Presumably a generalization of the Roam/Logseq idea.
more emphasis on "properties", a block is text plus whatever
Relationships are richer
Blocks have an identity and stable id
No pages at fundamental level
A page-block is a block with an ordered list of children
A Page is a page-block with a title and some child relationships
That is not quite the right name, but what I want: the ability to readily create sidenote comments. I mean Google Docs has this, so not rocket science.
2D sketching but only when you want it.
I want to be able to pour all my old journals and everything else and not have to worry
Presumably db size is not an issue, just have to make sure you don't do O(n) things.
Search might be tricky, might need to be weighted by date to not be overwhelmed.
Bottom up in a layered way that supports experimentation.
This could mean a lot of things, but one thing I just thought of: it should be possible to publish something that **includes procedures/code** and have it continue to work despite possible evolution of the infrastructure.
That seems rather hard and not really worth it, I don't have a use case, but that is kind of the point.