satellitecomponent / neurite Goto Github PK

Running it on kali linux and im having an issue utilizing the local model. the check box is tick on but when i click install nothing happens... it would be nice to have an acknowledgment or a progress bar...

Primary app runs fine, nice work -- but there may be an issue with the servers:

node start_servers.js
file:///mnt/repos/Neurite/Localhost%20Servers/start_servers.js:1
const { spawn } = require('child_process');
^

ReferenceError: require is not defined in ES module scope, you can use import instead
This file is being treated as an ES module because it has a '.js' file extension and '/mnt/repos/Neurite/package.json' contains "type": "module". To treat it as a CommonJS script, rename it to use the '.cjs' file extension.
at file:///mnt/repos/Neurite/Localhost%20Servers/start_servers.js:1:19
at ModuleJob.run (node:internal/modules/esm/module_job:194:25)

Node.js v18.15.0

The parsing for nodes and edges has been improved over time, but is still a major undertaking.

There is still room for further testing, validation, and potentially major efficiency improvements.

When using pinch to zoom on an iphone in safari, there are times when the rendering fully disappears. This occurs in some areas more than others, and the rendering can reappear from changing the position/zoom.

While I am about to implement a whole range of types of nodes within the text nodes themselves via syntax highlighting, refactoring the way in which nodes are defined has a number of downstream benefits.

While duplicate titles are avoided when entering titles from node headings, this is not the case when typing into the Zettelkasten.

Fixing this issue would ensure any duplicate Ai titles are handled properly.

The relevant files are in the Zettelkasten folder.

While all other aspects of Ai nodes are saved, restored codeblocks within Ai nodes currently miss all new lines. This does not actually effect code execution or Ai memory but would be nice to fix. The relevant code is in ainodes.js

Hi,

Thanks for building such a talented application. I love the idea of fractal workspace of text links.

When I was playing with Neurite, I met two times of data loss accidentally. I accidentally close the browser tab by clicking the close tab shortcut I set before first time. Then I try to explorer the local ai feature, and the app suddenly crashed. It make me frustrated and can't trust the app.

Can we auto save the data to local storage and let user load them if necessary?

Thanks.

Connect two or three AI nodes together, give them rules in how and when to reply to one another. Then connect them together and ask one a question. You get about 2 messages in before prompt error floods. Are they meant to be linkable this way or have I missed something?

A not yet implemented improvement is the ability for saved networks to retain their event listeners.

A reminder that Ai nodes are still at an early stage in their implementation and could require significant improvement and spawn many additions.

The main thing right now is to further test them, improve their safety/error handling, then we can move on to increasingly interesting additions.

It would take as long to explain all the potential ideas as it would to build them, so I would rather just get to the building part.

Right now, they already have a significant feature set, but are not fully reliable yet.

In the case of pasting notes into the zettelkasten, not all connections between nodes are recognized.

We use the processAll flag to enable the ZettelkastenrProcessor to handle all nodes as opposed to just the currently changed individual node. This allows for much more efficient input to individual nodes while still handling cases where all nodes need to be processed.

All connections between notes in the zettelkasten that are not found on paste can be restored by creating a node using SHIFT + DOUBLE CLICK in the fractal. This is another case where processAll is set to true. Handling the paste event is more troublesome. and does not seem to set the processAll flag correctly. One issue could be having to recognize the paste event manually in zetcodemirror.js.

The relevant code is in the zettelkasten folder found in the js folder. At the top of zetcodemirror.js and the entirety of zettelkasten.js.

While I have made a few attempts to get syntax highlighting within text nodes, they have not been sufficient enough solutions for release. Having syntax highlighting for reference tags and even code blocks within nodes would be a significant enhancement to Neurite.

The relevant files are in

Neurite\js\nodes\createnodes\nodedef.js
createnodes\contenteditable.js

One of @Paul-31415's fascinating contributions to Neurite is the ability for the fractal rendering to fully recalculate the svg positions at moments when a full refresh of the view is necessary. This is a significant reason why one can zoom so far into Neurite and could perhaps always have potential for both creative and/or obvious further iterations.

The relevant code is found in mandelbrot.js

Here is a snippet.

function recalc_svg(oldSVGpan,oldSVGzoom) {
    let node = svg_bg;
    for (let c of node.children){
        let path = c.getAttribute("d");
        let parts = path.split(/[, ]+/g);
        let coord = 0;
        let r = [];
        for (let p of parts){
            if (p.length && !isNaN(Number(p))){
                let c = coord?'y':'x';
                p = Number(p)/oldSVGzoom + oldSVGpan[c];
                p = (p-SVGpan[c])*SVGzoom;
                coord = 1-coord;
            }
            r.push(p);
        }
        c.setAttribute("d",r.join(" "));
        c.setAttribute("stroke-width",c.getAttribute("stroke-width")*SVGzoom/oldSVGzoom);
    }
}

This area of the code is the most directly relevant but is also not the full picture.

The current version of the recalculation is vastly improved from its initial

// TODO

however, one issue is a flickering where it appears the view is rotated for a frame before correctly restoring.

This is an issue which may not even be that hard to fix compared to what we have tackled so far, but is an area of the code I wanted to point out, as it is an entry-point into thinking about how the fractal rendering methods could be further expanded upon.