3 ms·

Very convenient of Mr. Willison to omit the fact that Bun's upstream changes are total garbage and would not be upstreamed regardless of any policies, omitting

Very convenient of Mr. Willison to omit the fact that Bun's upstream changes are total garbage and would not be upstreamed regardless of any policies, omitting LLM generated code or not, since they are, as a zig core team member articulated in a classier way, shite.

throwa356262 6mo ago
Also, that zig team is already working on other approaches that are better and more stable than what Bun team did: https://ziggit.dev/t/bun-s-zig-fork-got-4x-faster-compilation-times/15183/19 https://ziggit.dev/t/bun-s-zig-fork-got-4x-faster-compilatio...
- 000ooo000 6mo ago
  Notable quotes: >There’s the 4x speedup claimed by the Bun team, already available on Zig 0.16.0! >Each [incremental] update is taking less than 0.4s, compared to the 120+ seconds taken to rebuild with LLVM. In other words, incremental updates are over 300 times faster on this codebase than fresh LLVM builds are. In comparison, an enhancement capped at a 4x improvement is pretty abysmal. [..] Again, this feature is available in Zig 0.16.0—you can use it!
fg137 6mo ago
I have learned to take always Willison's words with a giant grain of salt, despite how popular those articles are here.
- simonw 6mo ago
  How can I do better?
  - n42 6mo ago
    Quality over quantity
    - simonw 6mo ago
      I aim for both. My blog is a combination of different content types. "Entries" are the ones I spend the most time on - https://simonwillison.net/entries/ https://simonwillison.net/entries/ Links and notes are more short form - I try to keep the quality high (especially with regards to accuracy) but they're also much higher volume than entries: https://simonwillison.net/blogmarks/ https://simonwillison.net/blogmarks/ and https://simonwillison.net/notes/ https://simonwillison.net/notes/
  - fg137 6mo ago
    More critical thinking. Many of your articles just go along with whatever another company/person says without questioning it, to the point that I don't find those articles valuable at all. I don't think I am alone, based on criticism in the HN discussions under those articles. And I don't expect you or anyone to be an investigative journalist, but there is a difference between writing blog articles vs being a megaphone.
    - simonw 6mo ago
      That may be true for many of the short form pieces - there's only so much you can say in a couple of paragraphs - but I try to inject a whole lot of additional commentary, often critical, in my longer form work. Recent examples from https://simonwillison.net/entries/ https://simonwillison.net/entries/ - https://simonwillison.net/2026/Apr/27/now-deceased-agi-clause/ https://simonwillison.net/2026/Apr/27/now-deceased-agi-claus... pulled apart the absurd deal between OpenAI and Microsoft where OpenAI could declare "AGI" and Microsoft would lose their rights to that IP - https://simonwillison.net/2026/Apr/23/gpt-5-5/ https://simonwillison.net/2026/Apr/23/gpt-5-5/ pointed out that, while OpenAI had said that GPT-5.5 was not yet available via their API for safety reasons, they also recently blessed accessing their unofficial Codex-only API as part of their attempt to tempt OpenClaw customers away from Anthropic... which meant they DID offer and approve of an API for GPT-5.5 after all - https://simonwillison.net/2026/Apr/22/claude-code-confusion/ https://simonwillison.net/2026/Apr/22/claude-code-confusion/ was highly critical of Anthropic's decision to A/B test the pricing of Claude Code, not to mention the messy way they had done that. - https://simonwillison.net/2026/Apr/21/gpt-image-2/ https://simonwillison.net/2026/Apr/21/gpt-image-2/ pointed out how much worse the previously celebrated Nano Banana is as at image generation compared to ChatGPT Images 2.0, while also showing ChatGPT Images 2.0 will cheat and add a raccoon if you ask it to solve its own puzzle. - https://simonwillison.net/2026/Apr/18/opus-system-prompt/ https://simonwillison.net/2026/Apr/18/opus-system-prompt/ was the first coverage I saw anywhere of the new eating disorder section of the Claude system prompt. I also repeated my criticism there of Anthropic's decision not to publish their tool descriptions even when they claim to have an openly documented system prompt. - https://simonwillison.net/2026/Apr/16/qwen-beats-opus/ https://simonwillison.net/2026/Apr/16/qwen-beats-opus/ demonstrated the new Opus 4.7 drawing a worse pelican than an open weights model I can run on my laptop. (Though to be fair I manly concluded this as evidence that my own "benchmark" is of very limited value.) All of those examples are from the past three weeks. I don't think I'm an uncritical megaphone for the messages these companies want to convey.
simonw 6mo ago
I hadn't see that post when I wrote this. I've updated it now to add a link. What were you trying to imply by "very convenient"?

throwa356262 6mo ago

Also, that zig team is already working on other approaches that are better and more stable than what Bun team did: https://ziggit.dev/t/bun-s-zig-fork-got-4x-faster-compilation-times/15183/19 https://ziggit.dev/t/bun-s-zig-fork-got-4x-faster-compilatio...

000ooo000 6mo ago
Notable quotes: >There’s the 4x speedup claimed by the Bun team, already available on Zig 0.16.0! >Each [incremental] update is taking less than 0.4s, compared to the 120+ seconds taken to rebuild with LLVM. In other words, incremental updates are over 300 times faster on this codebase than fresh LLVM builds are. In comparison, an enhancement capped at a 4x improvement is pretty abysmal. [..] Again, this feature is available in Zig 0.16.0—you can use it!

fg137 6mo ago

I have learned to take always Willison's words with a giant grain of salt, despite how popular those articles are here.

simonw 6mo ago
How can I do better?
- n42 6mo ago
  Quality over quantity
  - simonw 6mo ago
    I aim for both. My blog is a combination of different content types. "Entries" are the ones I spend the most time on - https://simonwillison.net/entries/ https://simonwillison.net/entries/ Links and notes are more short form - I try to keep the quality high (especially with regards to accuracy) but they're also much higher volume than entries: https://simonwillison.net/blogmarks/ https://simonwillison.net/blogmarks/ and https://simonwillison.net/notes/ https://simonwillison.net/notes/
- fg137 6mo ago
  More critical thinking. Many of your articles just go along with whatever another company/person says without questioning it, to the point that I don't find those articles valuable at all. I don't think I am alone, based on criticism in the HN discussions under those articles. And I don't expect you or anyone to be an investigative journalist, but there is a difference between writing blog articles vs being a megaphone.
  - simonw 6mo ago
    That may be true for many of the short form pieces - there's only so much you can say in a couple of paragraphs - but I try to inject a whole lot of additional commentary, often critical, in my longer form work. Recent examples from https://simonwillison.net/entries/ https://simonwillison.net/entries/ - https://simonwillison.net/2026/Apr/27/now-deceased-agi-clause/ https://simonwillison.net/2026/Apr/27/now-deceased-agi-claus... pulled apart the absurd deal between OpenAI and Microsoft where OpenAI could declare "AGI" and Microsoft would lose their rights to that IP - https://simonwillison.net/2026/Apr/23/gpt-5-5/ https://simonwillison.net/2026/Apr/23/gpt-5-5/ pointed out that, while OpenAI had said that GPT-5.5 was not yet available via their API for safety reasons, they also recently blessed accessing their unofficial Codex-only API as part of their attempt to tempt OpenClaw customers away from Anthropic... which meant they DID offer and approve of an API for GPT-5.5 after all - https://simonwillison.net/2026/Apr/22/claude-code-confusion/ https://simonwillison.net/2026/Apr/22/claude-code-confusion/ was highly critical of Anthropic's decision to A/B test the pricing of Claude Code, not to mention the messy way they had done that. - https://simonwillison.net/2026/Apr/21/gpt-image-2/ https://simonwillison.net/2026/Apr/21/gpt-image-2/ pointed out how much worse the previously celebrated Nano Banana is as at image generation compared to ChatGPT Images 2.0, while also showing ChatGPT Images 2.0 will cheat and add a raccoon if you ask it to solve its own puzzle. - https://simonwillison.net/2026/Apr/18/opus-system-prompt/ https://simonwillison.net/2026/Apr/18/opus-system-prompt/ was the first coverage I saw anywhere of the new eating disorder section of the Claude system prompt. I also repeated my criticism there of Anthropic's decision not to publish their tool descriptions even when they claim to have an openly documented system prompt. - https://simonwillison.net/2026/Apr/16/qwen-beats-opus/ https://simonwillison.net/2026/Apr/16/qwen-beats-opus/ demonstrated the new Opus 4.7 drawing a worse pelican than an open weights model I can run on my laptop. (Though to be fair I manly concluded this as evidence that my own "benchmark" is of very limited value.) All of those examples are from the past three weeks. I don't think I'm an uncritical megaphone for the messages these companies want to convey.

simonw 6mo ago

I hadn't see that post when I wrote this. I've updated it now to add a link. What were you trying to imply by "very convenient"?