我将所有 npm 包缩小 5%的失败尝试

In 2022, the author had an idea to decrease npm package sizes by 5% using Zopfli, which is a gzip-compatible compressor that creates smaller files but is slower. npm packages are distributed as gzipped tarballs.
- Proof of concept: The author tested on their own npm packages and found a 6.2% reduction in size with Zopfli, and it was backwards compatible. They also published one module this way without issues.
On 2022-05-27, the author asked colleagues for feedback on compressing the entire npm registry with Zopfli. They got feedback that decompression time could be significant and npm's lockfile couldn't handle re-compressing existing packages easily.
On 2022-05-29, the author submitted an RFC "Improving tarball compression while maintaining backwards compatibility" with 708 words.
On 2022-06-01, the npm CLI people discussed the RFC and felt it was worth evaluating further but cautious. The author built a proof of concept web app and attended the next meeting.
On 2022-06-15, in the meeting, the author was wordy and unclear but made the point that npm packages could be 5% smaller with Zopfli at publish time. Questions like "Who benefits?" and concerns about WebAssembly, performance, and licensing were discussed. The author was tasked with some investigations.
On 2022-07-31, after research and discussion, the author posted a comment on the RFC with pros and cons. The pros were a 4.5% size reduction for the top 250 npm packages and backwards compatibility. The cons were difficulties in integrating Zopfli, slower publishing, and not applying retroactively. The author closed the RFC.
Lessons learned: The author learned about making proposals, verbal communication during meetings, and that obvious wins may have trade-offs. They are still compressing their own modules with Zopfli.