- From: Harald Alvestrand <harald@alvestrand.no>
- Date: Thu, 09 Jan 2014 16:11:34 +0100
- To: public-media-capture@w3.org
On 01/09/2014 04:04 PM, Jan-Ivar Bruaroey wrote:
> On 1/9/14 4:04 AM, Harald Alvestrand wrote:
>> So you're suggesting that the SDP line number should be replaced with
>> a statement "we RECOMMEND that the informative debug message gives
>> the line number in the SDP where the error occured"?
>
> Yes, I think that covers typical use just great.
>
>> It's an informative message. I'll strongly recommend against placing
>> more restrictions than that on the format.
>>
>> If the line number really is needed by programs, the result will be
>> code like:
>>
>> if (chrome & chrome-version < 3.44) {
>> match /SDP line number: \d+
>> } else if (chrome & chrome-version < 4.48) {
>> match /Description line \d+ contains/
>> } else if (firefox & version > 3.27) {
>> match /SDP line: \d+/
>> }
>>
>> and so on and so forth.
>
> It's not needed by programs, except programs that would need to do
> this anyway IMHO. I think we let this cat out of the bag when we said
> "SDP parsing" was an API. It's not. It's an emergency access panel to
> the implementation.
To which viewpoint I agree.....
>
> My point was quite specific: Line-number isn't enough to cover
> Sylvia's rather ambitious code anyway, which likely would look like
> this already to parse the error messages for clues, not to mention
> browser-specific SDP tricks that might need to be dealt with, so this
> seems par for the course. Who needs a silver spoon to eat out of
> trashcans?
But SDP, for all its faults, has a specified grammar. So an SDP parser
can parse any compliant SDP, and "line number" has a specific meaning.
No browser implementor governs the syntax of SDP.
My objection above is to encouraging browser-version-specific code.
Received on Thursday, 9 January 2014 15:11:59 UTC