Add initial support for JS SpeechSynthesisErrorEvent

[asurdej-comcast]

1. Add new type SpeechSynthesisErrorEvent and expose it to JS
2. Extend PlatformSpeechSynthesizer speakingErrorOccured with error type arg
3. Send proper JS event on error condiditon
4. Stick to using older SpeechSynthesisEvent for platforms
   that doesn't support ErrorEvent yeti (mac)

[lauromoura]

There's an upstream commit from March in WebKit/WebKit@c8a7a8619b280 that adds a SpeechSynthesisErrorEvent with a slightly different API. For example, in SpeechSynthesisErrorEvent.idl:

<     Conditional=SPEECH_SYNTHESIS
---
>     Conditional=SPEECH_SYNTHESIS,
>     Exposed=Window
45c31,33
<     readonly attribute SpeechSynthesisErrorCode  error;
---
>     constructor(DOMString type, SpeechSynthesisErrorEventInit eventInitDict);
> 
>     readonly attribute SpeechSynthesisErrorCode error;

Among other changes, the upstream version matches the spec at https://wicg.github.io/speech-api/#speechreco-error with an extra implementation-specific messageattribute. So I think it would be nice if we backported it.

I tried applying it but couldn't test properly, as I didn't manage to get speech synthesis enabled in my WPE build (desktop). Is there any pending patch I could use to test it? Or another environment (e.g. buildroot)?

[asurdej-comcast]

Hi, thanks for your replay. I was not aware of an upstream patch. Indeed my patch misses SpeechSynthesisErrorEventInit dict initializer for ErrorEvent that is present in upstream version. On the other hand the other implementation doesn't allow to trigger different error codes from platform implementation so effectively only peechSynthesisErrorCode::Canceled is used (unless I miss something in the code).
I believe it's reasonable to wait for upstream patch backport and then rebase rest of my changes. Do you have any idea when this could happen roughly?

For the testing, WPE doesn't have any default PlatformSpeechSynthesizer implementation so the code won't be able to compile probably after enabling speech synthesis option. We do have our own implementation based on platform specific stuff but this is very Comcast specific. Currently there is only some mac implementations but not sure if we are able to use this (error event code is generic so any platform would be fine to test) or the mock version just to see the flow...

[lauromoura]

Hi, thanks for your replay. I was not aware of an upstream patch. Indeed my patch misses SpeechSynthesisErrorEventInit dict initializer for ErrorEvent that is present in upstream version. On the other hand the other implementation doesn't allow to trigger different error codes from platform implementation so effectively only peechSynthesisErrorCode::Canceled is used (unless I miss something in the code). I believe it's reasonable to wait for upstream patch backport and then rebase rest of my changes. Do you have any idea when this could happen roughly?

Hi @asurdej-comcast sorry for the delay. It's on my list for this week.

For the testing, WPE doesn't have any default PlatformSpeechSynthesizer implementation so the code won't be able to compile probably after enabling speech synthesis option. We do have our own implementation based on platform specific stuff but this is very Comcast specific. Currently there is only some mac implementations but not sure if we are able to use this (error event code is generic so any platform would be fine to test) or the mock version just to see the flow...

I'm thinking of giving a try and adding some mocks just to see if things get compiled to make it easier to track failures.

[lauromoura]

I've put the WIP backport in https://github.com/lauromoura/WPEWebKit/tree/speech-error-event-backport

And the upstream speech synth stubs in https://github.com/lauromoura/webkit/tree/speech-synthesis-glib-stubs

This Monday I'll test both together.

[lauromoura]

Updated API PR in #854

[asurdej-comcast]

#854 has landed
Rebased version of this PR in #863
Closing