Should datacontentencoding maps to Content-Transfer-Encoding ?

[fabiojose]

Should datacontentencoding maps to Content-Transfer-Encoding in the HTTP Transport Binding?

There is no mention about that like we have with datacontenttype, which maps to Content-Type. It is maybe an issue specifically with the binary content mode.

• datacontentencoding: https://github.com/cloudevents/spec/blob/master/spec.md#datacontentencoding
• HTTP Transport Binding: https://github.com/cloudevents/spec/blob/master/http-transport-binding.md#13-content-modes

[duglin]

I suspect this might be slightly different though because in the switch from 'binary' to 'structured' mode your choice of encoding data may need to change drastically based on what your structured format allows in the data attribute's value character set. And visa-verse.

E.g. when going from structured to binary you may actually go from base64 encoded to no-encoding at all.

I wonder if it would make sense to say:

• datacontentencoding and HTTP's Content-Encoding headers serve similar purposes
• whether they are the same values in both structured and binary modes will vary depending on the structure serialization chosen

Dunno. What do other people think? Could just be more of an FYI kind of thing to help people understand why datacontentencoding exists

[duglin]

hmm is it Content-Encoding or Content-Transfer-Encoding?

[fabiojose]

Like we have in the spec, datacontentenconding is based on the Content-Transfer-Encoding spec.

Yes, is Content-Transfer-Encoding what I am asking.

Here: https://tools.ietf.org/html/rfc2045#section-6.8

[clemensv]

While the fields correspond directly, I wouldn't prescribe a mapping because it's more than a little weird to carry binary as Base64 where you can carry Binary raw. For an SMTP transport, I would prescribe a mapping, but not for any of the transports we've got.

[fabiojose]

Just to be clear for me, consider bellow.

If we have this event:

{
  "specversion": "0.3",
  "id": "45c83279-c8a1-4db6-a703-b3768db93887",
  "type": "com.github.pull.create",
  "source": "https://github.com/cloudevents/spec/pull/123",
  "time": "2019-06-21T17:31:00.000Z",
  "datacontenttype": "application/json",
  "datacontentencoding": "base64",
  "data": "eyJtdWNoIjoid293In0="
}

And map that event to binary, we end up with this? ce-datacontentencoding:base64

curl -X POST \   
     -d'eyJtdWNoIjoid293In0=' \
     -H'Content-Type:application/json' \
     -H'ce-specversion:0.3' \
     -H'ce-type:com.github.pull.create' \
     -H'ce-source:https://github.com/cloudevents/spec/pull/123' \
     -H'ce-id:45c83279-c8a1-4db6-a703-b3768db93887' \
     -H'ce-time:2019-06-21T17:31:00Z' \
     -H'ce-datacontentencoding:base64' \
     http://localhost:3000/v03

[duglin]

Shouldn't ce-datacontentencoding map to the equivalent HTTP head in the 2nd snippet?

[duglin]

Anyone want to comment on this one?

Should datacontentencoding maps to Content-Transfer-Encoding in the HTTP Transport Binding?

@clemensv @cneijenhuis

@fabiojose perhaps open a PR to force the discussion?

[fabiojose]

@duglin , yep! I will do that!

[evankanderson]

I think #484 closes this out, removing datacontentencoding as a CE attribute.

[duglin]

Closed due to #492 - if incorrect please let me know