W3C home > Mailing lists > Public > public-html-media@w3.org > May 2014

Re: [MSE] Inconsistent error reporting for appendBuffer/appendStream

From: Aaron Colwell <acolwell@google.com>
Date: Mon, 12 May 2014 05:26:38 -0700
Message-ID: <CAA0c1bCdP3TgiMV_7KqFQAhiE_x62VuQPoFUYjuzApfXuEx0Eg@mail.gmail.com>
To: "<public-html-media@w3.org>" <public-html-media@w3.org>
Hi Matthew,

I think an 'error' event should probably be fired at the SourceBuffer
whenever it triggers the 'end of stream algorithm runs with error ==
"decode"'. I think this would make it consistent with the appendStream()
errors. I'll add a step to the "decode" section of the "end of stream
algorithm" that runs the "append error algorithm" if updating is set to
true. How does that sound?

Aaron

On Thu, May 8, 2014 at 8:20 PM, Matthew Gregan <kinetik@flim.org> wrote:

> Hi,
>
> While looking at the error handling of various types of SourceBuffer
> appends, I noticed the observable events when an error occurs in
> appendBuffer is as follows:
>
> For buffer full error:
> - throw QUOTA_EXCEEDED_ERR
>
> For segment parser loop algorithm errors:
> - updatestart -> SourceBuffer
> - [endOfStream algorithm runs with error == "decode"]
> - sourceended -> MediaSource
> - error -> HTMLMediaElement (depending on element's readyState)
>
> And for appendStream is as follows, for stream read or buffer full errors:
> - updatestart -> SourceBuffer
> - error -> SourceBuffer
> - updateend -> SourceBuffer
>
> And the same as appendBuffer for segment parser loop algorithm errors.
>
> Is it intentional that only appendStream uses 3.5.3 Append Error Algorithm?
> The error event as described in the event summary doesn't seem to be
> specific to streams.
>
> Thanks,
> -mjg
>
>
Received on Monday, 12 May 2014 12:27:06 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:33:03 UTC