Jeff King

> Ideally, something like this would somehow be hooked into the post-commit hook (and possibly post-applypatch, post-receive, ... what's the full list that would be needed?) rather than polling every...

> I'd prefer something more portable. [than inotify] Fair enough. The nice thing about `inotify` (or similar systems) is that you don't have need support from git at all. The...

> Counting on your expertise and hoping that you're still in some way affiliated with GitHub, and also referring to [your comment](https://github.com/Homebrew/homebrew-core/issues/18044#issuecomment-329301763) regarding byte-stable tarballs, can you tell us... >...

> checksums.contentSha256 would be a more bulletproof hash of the contents of each file sorted in a stable manner. This should mostly work, but I'll mention one case where it...

It's not clear to me what the best path forward is. There _is_ cryptographic integrity. But with lots of caveats related to sha1, and how collisions work, and what attacks...

I care a lot less about the blog posts. I think their future is to be destroyed with :fire:. Any useful content there made it into the second edition of...

To be honest, I'm not sure anything needs to be done here. Yes, SHA-1 has problems, and they're not likely to get better. But the best known attacks aren't viable...

I suspect the problem is actually in the search code, not the book importer. In the model for a book section, we index the content and provide an `id` field:...

> So just confirming this is not just a "correct the URL on the page issue"? Right. The problem is that the URL is not found in this repository at...

> This doesn't seem a caching issue as reported in #1153 right @peff ? Right, it's definitely not caching. We just don't generate those pages (and doing so is non-trivial;...