• Notice regarding image hosts and attachments. See here.

What are you coding?

Created my own issue tracking system this afternoon as for me, the process of development has fundamentally changed, in that I don't need to make big roadmaps and spend lots of time estimating tasks, because ANY engineering task can be done in a matter of hours. The priority for development these days is keeping track of what you have actually done, and finding ways to improve.

With that in mind I made a Rust-based server that presents a web interface and MCP, as well as claude code and opencode integrations. It has a CLI that the CC and OC integrations use.

As well as tracking what work got done, and why, it keeps records of the LLM sessions, number of turns, context size, and uses an LLM to summarize the session into the item description. It also tracks tokens used and estimates the real API pricing cost of the work...which is quite important for future planning.

There's a concept of 'loose ends' which is a catch all for minor work that comes up while building features, little extra enhancements or fixes.

Added discord webhooks support and a few other bits and pieces.

It's very hard to justify paying for tools like this for personal or small projects now, when you can so easily just create your own. Very liberating.
 
I've been using appsheet for years for small apps like vehicle checks at work, and just started vibecoding some new work planners. Looks fantastic, but I'm not convinced with the security aspect of it tbh!
 
I've been using appsheet for years for small apps like vehicle checks at work, and just started vibecoding some new work planners. Looks fantastic, but I'm not convinced with the security aspect of it tbh!

Yeah. If you're vibecoding you want security from something other than the code produced. I run the code locally hosted in DDEV. It can't be accessed outside my PC so security isn't a concern.
 
Here's the first video drop of the command and control, tactical planning, etc. prototype application and framework I mentioned a few posts back. It's a little rough around the edges but it's a start. For something quite basic as it stands, there is a lot going on under the hood! :cry:

 
I've always had an interest with coding but never really had the time to sink hours in learning to finish projects off. However, this year I've been going a lot of vibecoding for apps for my use. My favourite two so far are an internal job tracking app and an app to pull stock information from a supplier, allows me to edit the description, see buying price and my selling price and then pushes it out to various sites.

There's loads of job tracking apps in my field, but most of them didn't fit my workflow or had too many features that I didn't need and made it look messy. Any features I did need were paid addons.
 
I've gone a bit out of control now and have built my own IDE for working on my unreal projects. Very much LLM-first, but stripped back to just what I need, viewing files, building and launching editor sessions.

Built in Rust.

Really is a paradigm shift in software development, being able to roll your own tools this easily.
 
Waiting for Python 3.15 to be released tomorrow before I start a new project. No idea if it is going to be successful but I'm quite motivated to get my hands dirty.
 
IDE is now fully working, with Opencode and Claude Code. File browsers, Subversion and Git support, but primarily aimed at unreal projects. Neat.

Uses 42mb of memory :)
 

Attachments

  • tethys-screen.webp
    tethys-screen.webp
    79.8 KB · Views: 26
I have just changes my audio app's main control GUI from Win2d to SkiaSharp to test how it is, don't know if Win2d is still getting updated. It was not much work and it runs much better than I expected considering its a software render(It does have hardware support, not looked at that yet). Memory usage is normally between 150MB-14GB depending on the project;)
 
Gemini helped me update my smart alarm project I built in Lockdown on an ESP32. Very happy with the end result.
 
I have just changes my audio app's main control GUI from Win2d to SkiaSharp to test how it is, don't know if Win2d is still getting updated. It was not much work and it runs much better than I expected considering its a software render(It does have hardware support, not looked at that yet). Memory usage is normally between 150MB-14GB depending on the project;)

Claude's pretty good for optimising code to run on specific hardware. That enigma code breaking went from 8 hours in Python, to 2 minutes in C++/SIMD, and it could have been faster still in GPU.
 
Claude's pretty good for optimising code to run on specific hardware. That enigma code breaking went from 8 hours in Python, to 2 minutes in C++/SIMD, and it could have been faster still in GPU.
I have already multi threaded and vectorised most of the audio render code. The app uses lots of RAM because I cache the tracks/samples and sound fonts are large.
 
What SharkyUK is doing is a little different - while a similar presentation style as it stands GEV is more just the front end to what SharkyUK is doing - though both would do similar depending on the API feeding into them and level of interactivity with that source data.

Thanks - I couldn't have put it better myself! :)

IDE is now fully working, with Opencode and Claude Code. File browsers, Subversion and Git support, but primarily aimed at unreal projects. Neat.

Uses 42mb of memory :)

Looks cool. What is the underlying tech stack you are using?
 
Looks cool. What is the underlying tech stack you are using?
Rust and gui-kit and gpui-component for the UI. Alacritty for the terminal, a couple other libraries for file parsing.

Since last post I've added a basic debugger, can attach, set breakpoints, view stack trace. This afternoon I've also added an automation test runner that talks directly to UE editor over UDP for running tests. Vertical tabs support.

Been daily driving it for a few days now, not opened Visual Studio since :)

Got quite a lot of little cosmetic enhancements to add, around general UI panel behaviour etc, saving layouts and whatnot.
 
I have a finalised version of the video I posted previously showing my prototype command and control app/framework; I just tarted it up with an intro and some captions. :cry:


I have refined the UI (and will continue to do so as more features are added). For example, "Layers" are now available. This means that symbols, overlays, etc. can be placed on different layers on top of the map/terrain and be enabled or disabled, or their rendering order changed. The rendering system implemented this layered behaviour already but I hadn't wired it through to the UI.

There is also a full search and filter capability for entities now, too; although this is not shown in any of the screenshots.

55587289169_e8762a2bb4_o.jpg


I have also added a search capability that allows the user to immediately jump to a known/named place. This is limited to the UK for now but the mechanism can be extended globally. There are approximately 50,000 UK geolocated places in this prototype - city names, towns, roads, that sort of thing. It is just a database of names and locations so easily extendable.

55587098961_01c68ed148_o.jpg


Overlay support has been added - although restricted to rings, cylinders and domes at present (representing things like sensor range, operational ceilings, etc,) with bearing and sweep modifiers. User created tactical overlays are to follow soon. Overlays can exist on different layers and, depending on their type, can drape over the underlying map/terrain or show their 3D boundaries.

55587098956_693243257d_o.jpg


55586066617_365a0e8cd5_o.jpg


55587098981_acd9aec6da_o.jpg


55587289194_d7ffb3ce1e_o.jpg


55587215333_a88ceda0a9_o.jpg


55587215368_f7f697891b_o.jpg


They (the overlays) look a bit more impressive in action than in static screenshots as they all move/adapt in real-time and are fully editable.

I'm currently working on a line-of-sight feature, entity laydown, improved editing capabilities, undo/redo, and have lots of other things planned too. This could turn into quite a project actually. :cry:
 
I have dropped another video about how I render the Earth at scale as a couple of folks were asking about it elsewhere.


I have also started implementing a cloud layer system that will, eventually, be realised using volumetric rendering. For an initial pass, I decided to keep things simple and cheap so opted for a layered shell that wraps the planet, onto which a 2D cloud layer is projected. The cloud data is randomly generated at app startup using 5 octaves of Perlin noise and this can be used to generate a few clouds to completely overcast. Rather than suffer the cost of randomly seeding and generating a new cloud layer on demand, the randomisation of the cloud layer acts as an offset into the one-time generated data set instead hence the performance hit is negligible. The clouds also cast ground shadows and have absorption, scattering and density properties (to give a faux 3D effect despite being 2D in nature). It works better than I expected so will stick with it for a while. I can control the height of the cloud layer above mean sea level too. Of course, it breaks down a bit if the camera nears the height of the cloud layer!

55602032366_841e57fee3_o.jpg


55602434080_071c54c870_o.jpg


55601013547_8999b2e172_o.jpg


55602230569_12197925ec_o.jpg
 
Last edited:
Back
Top Bottom