Buckets and files

The Buckets tab lists what the store holds; a bucket opens as a table of files and folders with a filter, an upload button and a breadcrumb.

Buckets

Created and deleted from the Buckets tab, for a login that may. A name S3 itself would reject is refused where it is typed. A pinned store shows its one bucket and lists nothing else.

Files

• Upload streams through the daemon into the bucket; there is no size held in memory.
• Download streams back the same way, through GET …/buckets/{bucket}/download?key=.
• Delete removes one object.

A key is not a path: ../ means nothing to S3 and would name an object called that. Cubeship refuses it anyway, because its own listing is built from prefixes and a name that walks out of the folder it was uploaded to is a file that appears somewhere nobody put it.

What a folder is

S3 has no directories. A listing returns keys and common prefixes, and a folder is the second of those: it exists while something is under it and stops existing when the last thing goes. Three things follow:

• Deleting a folder deletes everything under it, at every depth, because there is nothing else to delete. The dialog says so.
• An empty folder is a zero-byte object whose key ends in / — the convention every console uses. The marker is hidden from the listing of the folder itself.
• A folder has no size. Working one out is a listing of everything under it, on every row.

Roles

All of it is an admin's: a bucket is data, and reading data is the line a member does not cross. Through the API it is under /objectstores/{name}/buckets.