Re: SpeechRecognitionError use of enum instead of const

Dominic,
I agree that it is better to omit "other", and as you say: if a new error
is needed later I think it'd make far more sense for a user agent to return
a specific error code with a vendor prefix (such as
"webkit-user-cancelled") rather than "other".

One reason for this is that the javascript code can (and should) check for
all the specific errors that it can handle, and if it's not one of them,
provide some default handling for any it doesn't handle (including
"other"). Said another way, code that checks for "other" is going to break
if there's additional error codes added in the future.

So I agree, I propose removing the error code "other". If there's no
disagreement, I'll update the spec with this on Wednesday.

/Glen Shires


On Mon, Oct 8, 2012 at 10:17 AM, Glen Shires <gshires@google.com> wrote:

> I've updated the spec with the that change: specifically: Changed
> SpeechRecognitionError to use enum instead of const.
> https://dvcs.w3.org/hg/speech-api/rev/e4e3ab962f1d
>
>  As always, the current draft spec is at:
> http://dvcs.w3.org/hg/speech-api/raw-file/tip/speechapi.html
>
>
> On Mon, Oct 8, 2012 at 8:42 AM, Dominic Mazzoni <dmazzoni@google.com>wrote:
>
>> I'm assuming this is for all enums throughout the spec?
>>
>> Also, I haven't been commenting on the speech reco part of the spec, but
>> I'm a little concerned about the "other" error enum. Is there a need to
>> actually have that in the spec? I think it's assumed that a well-written
>> client would have to be prepared for errors outside the spec (for future
>> expansion), and if a new error is needed later I think it'd make far more
>> sense for a user agent to return a specific error code with a vendor prefix
>> ("webkit-user-cancelled") rather than "other".
>>
>> - Dominic
>>
>>
>> On Mon, Oct 8, 2012 at 8:36 AM, Hans Wennborg <hwennborg@google.com>wrote:
>>
>>> On Fri, Oct 5, 2012 at 10:34 PM, Glen Shires <gshires@google.com> wrote:
>>> > My understanding is that the use of const is discouraged in favor of
>>> strings
>>> > or enumerations: http://dev.w3.org/2006/webapi/WebIDL/#idl-constants
>>> >
>>> > Based on this, I propose changing our SpeechRecognitionError from
>>> "const" to
>>> > the following "enum" (no change to the corresponding definitions). If
>>> > there's no disagreement, I'll update the spec with this on Monday.
>>> >
>>> >     interface SpeechRecognitionError : Event {
>>> >         enum ErrorCode {
>>> >           "other",
>>> >           "no-speech",
>>> >           "aborted",
>>> >           "audio-capture",
>>> >           "network",
>>> >           "not-allowed",
>>> >           "service-not-allowed",
>>> >           "bad-grammar",
>>> >           "language-not-supported"
>>> >         };
>>> >
>>> >         readonly attribute ErrorCode error;
>>> >         readonly attribute DOMString message;
>>> >     };
>>>
>>> Sounds good to me.
>>>
>>> Thanks,
>>> Hans
>>>
>>>
>>
>