Register MIME Types

[tparisi]

Just a task, we should do it soon, register MIME types for glTF and binary glTF

Current thinking is

model/gltf+json
modle/gltf+binary

[tparisi]

Related to #400

[RemiArnaud]

Not sure you got my previous message.
COLLADA Mime type is registered with IANA under model/vnd.collada+xml

it is IANA that insisted that we use vnd. prefix, considering Khronos as a vendor.

so, why not keeping the same for gltf?
model/vnd.gltf+json

I don't see the use of binary in mime type. octet-stream is used to designate raw binary formats.
so it would be
model/vnd.gltf+octet-stream

Note the way I understand '+' in mime type is not to decorate a type, but to provide alternate ways to handle it. I'm not sure one can use gltf+json and gltf+octet-stream to specify 2 different formats.

model/ is defined to be a 3D model. Would a 'model/vnd.gltf+octet-stream' or would it be the binary data that is used by the 'model/vnd.gltf+json'?

If it is the later, then I'd just use the standard application/octet-stream for the binary data. Meaning its just a bunch of bytes that can't really be used without more context.

[mlimper]

model/ is defined to be a 3D model. Would a 'model/vnd.gltf+octet-stream' or would it be the binary data that is used by the 'model/vnd.gltf+json'?

I think the idea was to have an own mimetype for the binary version of glTF. Since binary_glTF can be self-contained (binary header, JSON and binary data all in one file), the server would then send a binary_glTF file and communicate this using the respective mimetype. This would justify two different mimetypes for glTF, I guess?

[tparisi]

@RemiArnaud didn't realize IANA forced this... that's weird. Khronos isn't a vendor, it's a consortium. Well, we can try.

Regarding the second point it sounds like we'd want to go with model/gltf-binary (or with vnd. prefix), right? the dash is a legal part of the identifier, right?

[RemiArnaud]

I see 'dot' used more than '-'

so maybe this:

model/vnd.gltf.binary
model/vnd.gltf+json

[tparisi]

yeah sure that would be fine, too

[DRx3D]

tl;dr
Skip to bottom for a summary.

When Web3D registered the X3D MIME types, we kept meticulous records of the submission and resources. Most of the material is available at http://www.web3d.org/wiki/index.php/X3D_MIME-Type

A standards tree registration requires submission by a standards-related organization. I do not know Khronos qualifies in the eyes of IANA and needs to have adequate documentation. ISO is a recognized standards-related organization as seen by Web3D's submission and approval of model/x3d.

The MIME format is {type-name}/[facet-name.]{sub-type name}[+/-] [; parameters]

Model does appear to the proper {type-name}

If allowed, the {sub-type name| would be 'gltf'. Only if required would the facet-name be used - probably vnd. It is much preferable to get to use the first one. The dot character has a specific meaning (delineator) and is only allowed with a facet (e.g., 'vnd').

At the time Web3D was registering X3D, we were told that only "approved" suffixes would use +{suffix} (e.g., +xml). The use of '-' is as a word separator (e.g., 'x3d-vrml')

As far as I can tell any formatting suffix used is equivalent to application/{suffix}. So for '+json' the entire file would need to follow the JSON encoding rules.

As I understand gltf (which is by no means a complete understanding), the 'parameters' field might be a better indicator for binary vs. json, especially if the file is not fully JSON.

There are also restrictions on standards tree (non vnd) MIME types in both the copyright of the standard document and proprietary formats.

Summary:

• Actual registration process is fairly easy
• Intellectual property issues may need to be addressed
• Format is one of (or some near variant)
  • model/gltf
  • model/vnd.gltf
  • model/vnd.gltf+json
  • model/vnd.gltf-json
  • model/vnd.gltf-binary
• The process takes a month or two per round, probably 2 rounds

[pjcozzi]

@DRx3D thanks for the detailed explanation. glTF is a JSON file. It might reference external files such as images, but those have separate MIME types. "Binary glTF" is a glTF extension that defines a binary container format, which embeds the JSON and binary data (geometry, images, animations). Based on this, do you think we should propose:

• model/gltf+json // standard glTF
• modle/gltf+binary // Binary glTF

[DRx3D]

Standard glTF (model/gltf+json) is correct (if it goes standard)
There is no generic binary subtype. You would either need to use model/gltf-binary or fit it into one of the recognized sub-types.

[tparisi]

@DRx3D yes that works for us, gtlf-binary. Thanks so much...

[pjcozzi]

OK, so we are initially proposing:

• model/gltf+json
• model/gltf-binary

[RemiArnaud]

yes, exactly what I already wrote above about binary.

Also, you should use vnd, because during the discussion between Khronos and IANA about MIME type, IANA asked to use vnd. Also DrX3D explained why.

Should be:
model/vnd.gltf+json
model/vnd.gltf-binary

[emackey]

For what it's worth, Cesium is currently using:

model/vnd.gltf+json
model/vnd.gltf.binary

[RemiArnaud]

then there's no need to change IMHO -> what Cesium is currently using is fine.

[tparisi]

IMHO we shouldn't have to have a vnd. prefix in the MIME type. Khronos is not a vendor, it's a non-profit standards body. Shouldn't have happened with COLLADA either. So I would like to push on this.

[RemiArnaud]

You may want to contact Neil or Mark Barnes for the trail of emails between IANA and Khronos.
Anyways, someone from the Khronos Group need to contact IANA again for gltf mime-type if you want to drop the vnd. Would be nice to drop vnd for collada as well. Maybe by now IANA has heard of Khronos involvement in the web space :-)

[tparisi]

yes that would be good, too. @pjcozzi who is going to talk to IANA about this for us?

[DRx3D]

The IESG (IETF's Steering Group) is the organization that determines which standards groups can submit, "Requests from other organizations will be confirmed by the IESG"
https://www.iana.org/assignments/iesg-recognized-organizations/iesg-recognized-organizations.xhtml

I quickly checked the IESG page (https://www.ietf.org/iesg/) and did not find an obvious page that describes the review process.

[pjcozzi]

@pjcozzi who is going to talk to IANA about this for us?

Looking into it.

[pjcozzi]

Labeling as post 1.0. However, I'd like us to get rolling on this right soon.

[uklumpp]

You can assign this to me if you want.

[pjcozzi]

@uklumpp I don't think you have the right GitHub permissions for this (email me if you want them), but consider this assigned. Thanks!

[pjcozzi]

Registered, big thanks to @uklumpp!

https://github.com/KhronosGroup/glTF/blob/master/specification/README.md#mimetypes