Duplicated mimetype info when sending an image to a room

When I send an image to an e2e room using Element Web, the clear event contains undocumented fields:

{
  "type": "m.room.message",
  "content": {
    "body": "image.png",
    "info": {
      "size": 16144,
      "mimetype": "image/png",
      "thumbnail_info": {
        "w": 166,
        "h": 72,
        "mimetype": "image/png",
        "size": 4770
      },
      "w": 166,
      "h": 72,
      "thumbnail_file": {
        "v": "v2",
        "key": {
          "alg": "A256CTR",
          "ext": true,
          "k": "rpj1sOJ6wO2oy3R0TkmzMtlTfq8gw34DDjBEomEcFg4",
          "key_ops": [
            "encrypt",
            "decrypt"
          ],
          "kty": "oct"
        },
        "iv": "mTh8NqPZ4XgAAAAAAAAAAA",
        "hashes": {
          "sha256": "VehmjJbux6m63u/zWs2zZCElqqGgQ5pFuUNzmm1uovg"
        },
        "url": "mxc://matrix.org/pKlpKPCIpzUBNElxmaAvuwbh",
        "mimetype": "image/png" << This is not documented and redundant info
      }
    },
    "msgtype": "m.image",
    "file": {
      "v": "v2",
      "key": {
        "alg": "A256CTR",
        "ext": true,
        "k": "7ApOyV_yijdMcXWMBvxApSwc_Nkl39GBxhSY3GdZRNI",
        "key_ops": [
          "encrypt",
          "decrypt"
        ],
        "kty": "oct"
      },
      "iv": "eH/zzQF1HEEAAAAAAAAAAA",
      "hashes": {
        "sha256": "EPedp3jPqlPPS4OCInieN9u5X4KHpbGVzb/1LAmIH/g"
      },
      "url": "mxc://matrix.org/TXyRLVIClHHgxJHFwcBfNMLd",
      "mimetype": "image/png" << This is not documented and redundant info
    }
  }
}

According to https://matrix.org/docs/spec/client_server/r0.6.1#extensions-to-m-message-msgtypes, EncryptedFile.mimetype is not documented (but are present in the examples!), and the information it contains is already present at other place in this content.

For some reason, there is the same problem on Element Android, so this issue is probably more a question: why are we adding those fields?

[anoadragon453]

I have a sneaking suspicion that the example was just ripped out of Element Web’s /devtools at some point without checking the contents too closely…

If the field is truly redundant and not necessary, I don’t think there’s any reason to add it to the spec - and the example should be corrected. Element Web - if they want - will need to remove the fields and take care that nothing crucial that relies on it breaks.

If the field is actually useful, then someone should write an MSC for it. Additionally, the field existed before unstable prefixes were introduced, so Element Web shouldn’t need to switch to unstable prefixes in this case.

[anoadragon453]

Now that the spec has removed these fields (MSC2582), I don’t believe there are any more blockers to stop sending this information.