Hank Donnay

My proposed new schema (#1340) should be able to capture this information.

/fast-forward

I've run this through the fuzzer for a while, with no crashers being reported. fuzz: elapsed: 30m1s, execs: 121110401 (0/sec), new interesting: 51 (total: 388)

Re-drafting this until I figure out what to do about percent-encoding.

cc @giuseppe This `zstd:chunked` implementation is largely based on reverse-engineering the code in containers/storage. Is there more formal documentation or test fixtures that I can cross-reference against?

Note to self: using linux's [non-POSIX record locks](https://man7.org/linux/man-pages/man2/fcntl.2.html) and a file-reopening scheme would allow the `fs.SubFS` interface to be implemented directly, but would require some thought about how that interacts...

ah, okay. Well I feel like I've got a handle on it, so I can help with the documentation when the time comes. Clair can use this to reduce disk...

Decided to punt any `httpreader` usage to a future PR. The tricky things for that future PR will be: - New fetcher implementation - New Indexer interfaces (either on the...

This is also going to be easier to review commit-wise, sorry.

> Decided to punt any `httpreader` usage to a future PR. > > The tricky things for that future PR will be: > > * New fetcher implementation > >...