I copied the files to the new locations on S3…

so a symlink in

protoc -> protobuf

should do in binaries right?

I don’t actually know what the current relationship between binaries and S3 is. E.g., do we actually need the binaries repo any more? It seems like all we need to preserve is the scripts that generate the binaries? Or am I missing an important point here?

the scripts that generate the binaries

I doubt they can generate all the binaries. Even if they can, some things would take a long time to build. afaict s3 is the mirror of the repo

Yeah, there are a ton of symlink hacks. We heavily extend the s3 mirror internally and do the same thing