Delgan

The GP misquoted the article. They actually wrote:

> Shopify has been using LLMs to build software since 2021

GitHub Copilot integrated into VS Code started becoming available that year. It’s definitely not a coding agent, but it was certainly LLM-assisted programming.

I, too, was curious to see it in practice.

Here is the ticket opened by @retr0id: https://github.com/oven-sh/bun/issues/28030

And here is the swarm of bots / LLMs / agents that open, review and bikeshed the PR before it's closed by the stalebot: https://github.com/oven-sh/bun/pull/28031

It's hilarious. But also a little sad.

Alas, GitHub has been plagued by bugs and UX regressions year after year.

I reported a bug last year about being unable to quote code blocks. It's quite a basic yet fundamental feature, right? They acknowledged the bug and moved on. To this day, quoting a block of code is still broken [1].

They simply don't care. I suppose their attention is focused on other subjects...

Anyway, I kind of accepted the "enshitification" of things I used to like. Fortunately, in this case, we can still hack our way around using custom userscripts [2].

[1] https://imgur.com/a/github-bug-cant-quote-blocks-of-code-Z9O...

[2] https://github.com/orgs/community/discussions/192665#discuss...

I was baffled by the technical blog post they published on this subject, in 2023: https://github.blog/engineering/architecture-optimization/cr...

They completely redesigned the code navigation to load it dynamically using React. One of the regressions introduced caused the entries to be duplicated when searching for a word using Ctrl+F. Their solution to address this issue? Decompose the source character by character, assigning each individual character its own distinct HTML node...

Needless to say, in addition to abysmal performance (it took seconds to display on my laptop), this also caused new problems (such as the inability to search for certain composite emojis). The worst part is that they seemed proud of their hack, since they wrote a blog post about it.

There is also a linked Youtube channel.

https://www.youtube.com/watch?v=7gi7nMuyUc4

He looks real.

It reminds me of the multiplayer section from CodinGame website: https://www.codingame.com/multiplayer/bot-programming

Although there is no specific implementation for Battleship, but I believe that some users sometimes have the possibility to create and submit their own games.

One month ago, I spent hours testing dozens of apps for "gamified todo list". I could not find any that would perfectly suits my needs (Habitica is too gamified for example).

Your app looks very promising. Polished UI. Intuitive. Unlimited habits. Simple "task -> award" principle without added complexity layers.

Something I'm having trouble finding in the apps I've tried is the distinction between "habit" and "task". An habit is something I want to do multiple time in a week. While a task is a one-shot TODO that I would like to execute at some point when I'm not procrastinating. Once done, I don't need to repeat it, but I like to be rewarded.

Your application seems to be more habit oriented. Is the notion of a one-time task something you could eventually incorporate?

On the contrary, loop-and-a-half is intended to avoid repeating stuff outside and inside of the loop body.

    while true:
        x = stuff
        if not x:
            break

Isn't it what Numba does quite successfully for a subset of Python?

[1]: https://numba.pydata.org/