Notify NFCReaders of failures when trying to get a message from a tag?

[leonhsl]

See
https://developer.android.com/reference/android/nfc/tech/Ndef.html#getNdefMessage().
Once a tag is discovered, we call android.nfc.tech.Ndef.getNdefMessage() to try to get a message from it, which may

• return null if the tag is in the INITIALIZED state, i.e. formatted but empty
• throw exceptions: TagLostException, IOException, or FormatException

In such cases do we need to let all active NFCReaders be aware of these failures so that they can choose how to proceed next? i.e. dispatching some error events properly to NFCReaders.

Currently http://w3c.github.io/web-nfc/#the-nfc-reading-algorithm only describes the case that a ndef message has already been gotten from the tag, so does the implementation in Chromium.

[kenchris]

So I assume the problem we have here is that 1) is an empty formatted tag that doesn't even have the empty record?

[leonhsl]

Exactly.

[kenchris]

Could we return an empty record?

[riju]

For context, this came up in review. In chromium implementation, I do see a kNfcEmpytMessage used to throw an exception while pushing empty message to tag. We could use something similar while reading also ?

[kenchris]

For TagLostException, we can just avoid dispatching the event as no tag was fully read.

[kenchris]

We should probably do like Generic Sensor and have an onerror event handler as well, which somehow can handle the error cases

[leonhsl]

So, for the null ndef message case we dispatch a reading event with a ndef message containing just an empty record; for IOException and FormatException cases we dispatch an error event. Right?

Note: These events are dispatched to ALL active readers, without any message filtering for matching their reader options.

[kenchris]

Sounds good, but I think we might want to consider adding a "formatted" type, similarly to the "empty" type

[beaufortfrancois]

According to #367, reading non-formatted tags result in a onreading event with empty records array for NDEFMessage.
I think we can close this issue right?

[beaufortfrancois]

Ooops! We should not close it. Dispatching event errors in IOException and FormatException cases is not supported anymore as we got rid of NDEFErrorEvent.

[kenchris]

So you suggest we readd the error event for that? Or other solutions?

[beaufortfrancois]

I didn't find any other solution sadly that satisfies me but re-adding the NDEFError event.
@zolkis for thoughts as well.

[zolkis]

I didn't find any other solution sadly that satisfies me but re-adding the NDEFError event.

I think so, too.

[beaufortfrancois]

For info, I have a PR at #432 that attempts to fix this.