Rename the Origin Private File System to the Bucket File System

[a-sully]

Fixes #92

---

Preview | Diff

[a-sully]

cc @annevk @asutherland @inexorabletash @mkruisselbrink

[tomayac]

Since the publicized name in blogs etc. has been origin private file system, I wonder what we do "brand"-wise. Would it be fair to keep saying "origin private file system" in general, and then more concretely say that "the origin private file system can contain one or multiple bucket file systems"?

[a-sully]

I think it's fine to continue to refer to it publicly as the Origin Private File System, which at this point seems to have some name recognition and has even crept into interfaces such as sqlite3 WASM. But from the perspective of the spec, we probably should use the more specific term? Developers presumably look at MDN rather than the spec itself anyways, and MDN could have a note explaining the discrepancy

Alternatively we could just leave things as they are (or perhaps make the minimal change of saying "an OPFS" instead of "the OPFS", to imply that there could be multiple per origin)

"the origin private file system can contain one or multiple bucket file systems"

I wouldn't suggest framing it this way, since it makes it seem like an OPFS contains several file systems. An origin (or more accurately, a storage key) can contain several buckets, and therefore several file systems (one per bucket)

[annevk]

Looks good modulo nit.

[annevk]

We should keep the existing ID. (For the other <dfn> change I don't think it matters since we introduced it two days ago, but this is much more likely to be referenced already.)

And FWIW, in my opinion it's fine if the ID has a different name from what it identifies. It nicely shows the specification has evolved.

[a-sully]

Done. I was ready to update all the refs anyways (see WICG/file-system-access#416) but it is nice to see the evolution :)

[tomayac]

I think it's fine to continue to refer to it publicly as the Origin Private File System, which at this point seems to have some name recognition and has even crept into interfaces such as sqlite3 WASM. But from the perspective of the spec, we probably should use the more specific term? Developers presumably look at MDN rather than the spec itself anyways, and MDN could have a note explaining the discrepancy

FYI, @chrisdavidmills helps us with documenting this on MDN.

Alternatively we could just leave things as they are (or perhaps make the minimal change of saying "an OPFS" instead of "the OPFS", to imply that there could be multiple per origin)

This makes sense to me, and is a minimal change.

I wouldn't suggest framing it this way, since it makes it seem like an OPFS contains several file systems. An origin (or more accurately, a storage key) can contain several buckets, and therefore several file systems (one per bucket)

This may be a stupid question, but calling it Bucket File System sort of implies we have agreement on Storage Buckets (deep-link to the relevant OPFS section)?! Yet it seems like Mozilla's suggested "worth prototyping" position from mozilla/standards-positions#475 is stalled on WICG/storage-buckets#36, and WebKit's not published a position yet looking at WebKit/standards-positions#181.

[annevk]

No, Bucket here refers to Storage Standard terminology. Not the Storage Buckets APIs.

[tomayac]

No, Bucket here refers to Storage Standard terminology. Not the Storage Buckets APIs.

Thank you! I see, as per this definition, which the Storage Buckets API also references. It's just a bit confusing that there's also a StorageBucket.

[annevk]

Well, that's an API instance of that concept, so it's quite logical in a way.