In my personal experience, GitHub Copilot is an absolute pile of shit. Maybe it's just that I've used smaller models on it, but it just felt like the worst available harness, even behind Google Antigravity's.
Is there any redeeming quality to it nowadays? I'm curious what those eight hundred thousand lines of Rust actually do.
my experience has been great with copilot cli, but yeah there is some memory leaks here and there and slowness which i hope this rust port can help with
I use copilot as my daily driver and haven’t seen much difference with it and Claude code’s output.
I think your answer is you’ve been using weaker models, that’s comparing apples to oranges. Have to try the same model on both harnesses. Also have to remember that GHCP harness is model agnostic, Claude and OpenAI can tightly couple their harness to their model so they will always have a slight edge initially. Making agnostic harnesses is much harder to do imo and if successful is much more valuable to customers.
> a terminal UI (TUI) on top of an agent loop. ... TypeScript, using Node.js as the framework and V8 for the execution engine, with Ink and React for UI. That’s a respectable choice for a TUI application
Who would respect that???
> And for the needs of a console application, the performance implications in terms of startup, responsiveness, throughput, and memory consumption are also reasonable
Literally none those things are reasonable and are specific issues people complain about because their expectations formed by other TUI apps are broken with such an approach
> A project that would have taken a whole team of developers a year or two before agents
Or it would've been a project that took less time overall with the better designed stack from the beginning since it could avoid time spent on this rewrite and its pains
Those kinds of in-place intense rewrites or ports of production software , assisted by AI probably is going on in other companies as well. High-value, verifiable.
Too bad this article didn't get much attention. Not everything is applicable to others, but valuable lessons, even if part of it is AI writtern (doesn't feel like it)
6 comments
[ 0.17 ms ] story [ 19.2 ms ] threadIs there any redeeming quality to it nowadays? I'm curious what those eight hundred thousand lines of Rust actually do.
I think your answer is you’ve been using weaker models, that’s comparing apples to oranges. Have to try the same model on both harnesses. Also have to remember that GHCP harness is model agnostic, Claude and OpenAI can tightly couple their harness to their model so they will always have a slight edge initially. Making agnostic harnesses is much harder to do imo and if successful is much more valuable to customers.
Who would respect that???
> And for the needs of a console application, the performance implications in terms of startup, responsiveness, throughput, and memory consumption are also reasonable
Literally none those things are reasonable and are specific issues people complain about because their expectations formed by other TUI apps are broken with such an approach
> A project that would have taken a whole team of developers a year or two before agents
Or it would've been a project that took less time overall with the better designed stack from the beginning since it could avoid time spent on this rewrite and its pains