iThoughtsX Development Blog

I'm still considering the same directions, however, I'm currently working on the view menu for view zoom in and zoom out and fit to map.

The other item I'll work on before I get to the copy paste stuff will be I've noticed a problem in the as yet unpublished Tori Android app. The problem I'm seeing is on a large map, if you fiddle around with opening and closing nodes, then the map 'saves' And saving on a large map is actually quite laggy and seems to freeze the interface up, which is annoying.

I'm considering these directions next

If you want to influence the direction please email me: simon dot ryan at gmail dot com

I noticed an issue with very large maps, they seem performant enough scrolling around, but if I change something (open close a node say) then the autosave triggers and locks it up for multiple seconds. I’ll not fix this yet though, I think it is instead time for a beta release.

Also copy and paste into and out of the app doesnt work for any platform - this is a must have afaic. This might be next after the above fix.

Final tests for release have commenced!

Release 0.3.0 done!

Found and fixed some weird regressions to do with the way the main site is published

Fixed regressions. Also fixed some scrolling issues while I was at it. Now it looks ready for a beta release.

I’ll try to use it on my Android as much as I can over the next day and then do a release before I add any more features.

Next thing is probably the side panel to allow editing node attributes. This will allow users to set shapes and colours

Fixed the juddering issue with huge maps.

Now working on some root node colour regressions under MacOS Tauri

Pinch to zoom is working on Android! (But not on Chrome desktop or Tauri Desktop)

I've also added canvas inertia. Because I noticed that this stopped once I implemented the pinch to zoom and the new methods. I'm trying to set it up so that it roughly replicates the browser's natural scroll inertia.

I'm also now looking at some performance issues with really large maps. If I completely expand all the nodes, I'm seeing some slight juddering when I scroll. Hopefully, I can fix this issue, although it's probably okay as it only affect the appearance if a HUGE map is scrolled.

Panning and scrolling looks ok now, starting on pinch to zoom code. Once I'm happy I'll do another beta release on this.

Hmm, I have a partially working 'free' canvas that pans and scrolls with two fingers or mouse scrolls. However I realise now that I need to make sure this section of code is nice and isolated from the rest of the code. Currently it is still fairly 'incorporated' in the one and only layout style that I’ve implemented. Today I’m spending time ensuring that the code is agnostic to the layout. I havent started on pinch and zoom yet as there are still a few issues with drag and drop interfering with the panning code.

Ok, time to attempt free moving 'canvas'

Here are the steps I'm thinking

Finished tweaking my AI harness, so it can run for much longer. Also I’ve pretty much switched to Qwen3.8-27B for coding. It does a very good job and seems pretty good at reliably staying on track. For larger, more complex architectural things (such as the recent major refactor) I’m sticking to GLM-5.3 as the best open weight model for intelligence

Doing some work on my Agent13 AI harness at the moment.

The next thing I’m thinking of developing is added a free moving canvas feel. Currently the app is limited to the screen width and sometimes gos outside the viewable area for deeply nested maps. Time to fix this!

Refactoring is done. I'm sitting on the new code using it as much as possible before pushing a release.

Meanwhile, I'm considering getting the copy/paste working next

Ok, it looks good, releasing 0.2.2 now!

Doing some serious re-factoring and more shape fixes, hopefully the last for a while. After this is done, I’ll do another release.

Fixed the shape and appearance of the root node

Tuned node max-width and wrapping to match the original app's appearance

Code is going through final tests, I’ll do a beta release today when the CI is happy

All colours and shapes are done

Next I’ll work on 'auto' text colour since I notice the dark blue nodes with black text doesnt look so good

I now have the 'visual ci' environment running, this means the AI can 'see' how things render in the browser. This means the Agent can iterate for much longer and doesnt claim it is 'done' until it can visually verify the changes made.

BTW: I'm back to Qwen3.8 27B, I'm not convince about 'flash', also I'm waiting for GLM-5.3 weights to appear.

Qwen3.8-Flash-Next is having a go at setting up a nice CI environment for me.

I'm starting to work on 'shapes' now.

Moved all the development progress stuff in here to start a new blog page

Got autosave working (along with file race contention)

Got 'recent files' working

Released this new website. It is now running the same code as the beta release. In fact you can use this site to open and edit maps. When I’m happy it is working, I’ll enable PWA install from the front page.

I've been working on the code for this website. Its currently running an old python version of my early code. I'll shortly cut the backend over to the new beta code (with some tweaks to it so it can serve up maps like this code does)

I have the Tauri Android app running now, I'm using it on my phone and still finding a few bugs, so I'll hold off the release until I update this website.

Got website mode working (this is how you are seeing this) - reads itmz from disk and renders as readonly webpage