Trying to make agent clients on low end laptops like a Celeron 900, not a C++/GTK developer so was trying to do it with python, been brutal trying to build python wheels on the single core 900mhz cpu, might look into compiling on a stronger PC or just committing to C++/GTK. I was able to get it to work on an intel 270.
Python gets pretty heavy for old hardware. Once you hit your MVP you may want to look at compiled languages for better single-threaded perf.
Yeah I don't think I have a choice for the Celeron M 900MHz case. I mean it was cool for me to just let it run overnight to build the temporalio wheel but other deps have been failing so I think I'll commit and use C++/GTK.
I could look at cross compiling too for the wheel to be available. I also do want to get better at C++ so would be good to practice.
Edit: to be clear I failed to make the temporalio wheel, something about maturin/rust related dep.
I've been using golang for a lot of single-binary stuff. I rather enjoy the language and it is pretty simple to reason about.
C++ sounds like... fun? For various definitions of 'fun'. At least you'll have a compiler chain that can target the OS/HW combo you're running.
Do you have a blog or something that you post updates on?
This is technically a burner hn account but I have a github account. I'm kind of all over the place projects wise one of those start a dozen finish 1 type of people.
Golang to me is a lot easier to learn than Rust and I had fancied it before. Although it's one of those things if you don't use it often... lose it. C++ I use it some for my hobby robotics projects and most recently I was working with ESP-IDF/bluetooth stack and that was all C/C++. I am more of a scripted language person eg. JavaScript/Python are my main languages.
For the old computer talk above this is the project, can see I did make an agent client (although a poor one) using Tkinter since the ASUS 1005HA was able to compile the python wheels (although it took a few hours).
I'm also going to use a local LLM I have a 4070 so I can run like Llama-3-8B through llama.cpp
https://github.com/jdc-cunningham/agent-gui