Hacker News Reader

How is pulling dependent on github?

Git pulling isn't unique to github and it works over http or ssh?

A neat thing about GitHub is that every file on it can be accessed from URLs like https://raw.githubusercontent.com/simonw/llm-prices/refs/hea... which are served through a CDN with open CORS headers - which means any JavaScript application running anywhere can access them.

Demo: https://tools.simonwillison.net/cors-fetch?url=https%3A%2F%2...

That feature seems common to other git hosts / forges. For example, here's one of Dillo's files, from a few commits ago, from their cgit-based host

https://git.dillo-browser.org/dillo/plain/src/ui.cc?id=29a46...

That doesn't have open CORS headers: https://tools.simonwillison.net/cors-fetch?url=https%3A%2F%2...

It's also not being served via a caching CDN, which means I don't feel comfortable running anything automated against it as that might add load to the server that they aren't ready for.

It's less about pulling and more about tools like DeepWiki making the assumption that its inputs live in GitHub, so repository URLs are expected to be GH URLs as opposed to a URL to a git repository anywhere.

That being said, there's no reason for tools like it to have those constraints other than pushing users into an ecosystem they prefer (i.e. GitHub instead of other forges).

Reply on news.ycombinator.com