Jiahao XU

Quickinstall uses cargo-zigbuild to cross compile to glibc 2.17, perhaps that's the souce of the error?

Yes, however note that many tests sit in an unpublished workspace https://github.com/rust-lang/cc-rs/tree/main/dev-tools/cc-test so if you'd like to test it, we would also have to publish cc-test as well

https://docs.rs/async-compression/latest/src/async_compression/codec/bzip2/encoder.rs.html IIRC there isn't any early flushing in async-compression

https://docs.rs/async-compression/latest/src/async_compression/codec/bzip2/encoder.rs.html IIRC there isn't any early flushing in async-compression

@kamikaze I think your `process_objects` can be optimized, it currently downloads and compress sequentially while it can be parallelised. The code wait for the current download and tarball append to...

@kamikaze one of the key difference between your demo and bzip2 cli is the I/O part, fetching from network has more latency and less bandwidth than disk I/O, and your...

> since I've swapped BzEncoder to XzEncoder @kamikaze it might be that xz creates much smaller archive that takes much faster to upload, since it's likely the network bandwidth is...

> WARNING: this is very likely to result in the same crate being built many times in parallel if someone runs the cronjob manually on ci/locally We have a mechanism...

> would there be any value in including cargo-binstall's tree of fallback architectures here? Honestly, I think we should just build the crate for all supported target anyway, but I...

> Pushing the whole queue on does sound like an interesting idea though. Would allow us to get rid of the randomness I guess. Yeah, and we can "save" the...