Re: CHANGE: [Bug 18486] New: Let RTCSessionDescription take a Dictionary parameter

On Fri, Aug 24, 2012 at 7:32 PM, Harald Alvestrand <harald@alvestrand.no>wrote:

> On 08/24/2012 05:51 PM, Randell Jesup wrote:
>
>> On 8/24/2012 10:43 AM, Tommy Widenflycht (ᛏᚮᛘᛘᚤ) wrote:
>>
>>> It has been pointed out to me that the stringify algorithm is broken,
>>> especially for RTCSessionDescription since the sdp member most certainly
>>> contains newlines. Should had noticed that myself, doh.
>>>
>>> Also some clarification regarding exactly what the end result is need to
>>> be put in the specification.
>>> We had a discussion regarding if this was meant to be JSON or not.
>>>
>>> /Tommy
>>>
>>> On Wed, Aug 22, 2012 at 2:10 PM, Tommy Widenflycht (ᛏᚮᛘᛘᚤ)
>>> <tommyw@google.com <mailto:tommyw@google.com>> wrote:
>>>
>>>     I'm fine both with removing the stringifier and letting it create
>>>     "JS object strings" as long as everyone understands that it
>>>     isn't necessarily JSON compatible.
>>>
>>>     JSON.stringify(object) != (string)object
>>>
>>
>> So, we have 3 options:
>> 1) Use the current stringifier (with fixes for newlines, maybe quotes)
>> 2) Move it to valid JSON (newlines, quotes, parens, ?)
>> 3) drop the stringifier and use "JS object strings" (what's the impact of
>> this?
>>
> If we drop the stringifier, I think the result of trying to interpret the
> object as a string is "{ object object }", which is well defined, but kind
> of useless.
>
> I wouldn't mind dropping it until we find a good reason to standardize it;
> people who want a particular stringification can write their own
> stringifiers.
>
>
>

+1

Especially since 100% compliant JSON can be produced with just one line:
JSON.stringify(object).
Why add duplicate functionality?

/Tommy

-- 
Tommy Widenflycht | Senior Software Engineer | tommyw@google.com | +46
734162531
Google Sweden AB, Kungsbron 2, SE-11122 Stockholm, Sweden
Org. nr. 556656-6880