Aleksa Sarai

It looks like *a lot* of the time is spent in decompression (which is sadly single-threaded and is the same as stock `encoding/gzip`). Hashing does cost us a fair bit,...

Ideally, yes. But we could also take another look at switching just the decompression side to `zlib` (as we discussed in #225 and #233). Unfortunately in the meantime `vitess` have...

While we can (and should) add support for zstd -- the issue is that most other image builders aren't producing zstd images (and others' don't support zstd yet) so we're...

`oci` wouldn't be moved to internal. Most of `pkg` would likely be moved, as well as `third_party` (though honestly, `third_party` should be removed entirely). Regarding `mutate`, I do really want...

> ok, so is the "oci/**" tag incorrect then? Yup, we didn't have a `go-api` tag -- I've added one now. > The API is mostly fine except that it...

@dependabot rebase

As an aside, it looks like `copy_file_range(COPY_FR_DEDUP)` [wasn't merged](https://lwn.net/Articles/659523/). But you can use `ioctl(FICLONERANGE)` or `ioctl(FIDEDUPERANGE)` (depending on which is the most correct way of doing it -- I think...

`FICLONERANGE` needs to be block-aligned (unsurprisingly) but unfortunately the block alignment is for *both* source and destination. This means that if we have different block sizes which are out-of-alignment of...

Some things that should be tested and discussed: * How bad is the Merkle tree hit? Should each individual file be linked from a map (or a packfile) of some...

I will definitely look into this, though it should be noted (and I think we discussed this in-person in London) that while it is very important for fixed chunking parameters...