[Matroska-devel] EBML data type constraints
dave at dericed.com
Tue Jun 23 19:52:11 CEST 2015
> On Jun 23, 2015, at 12:12 PM, wm4 <nfxjfg at googlemail.com> wrote:
> On Tue, 23 Jun 2015 08:44:15 -0700
> Ralph Giles <giles at thaumas.net> wrote:
>> On 2015-06-23 6:30 AM, wm4 wrote:
>>>> Is a 0 sized (un)signed integer allowed?
>>> I'd vote no. Didn't we discuss this already and concluded they're not
>>> allowed? But maybe I don't remember correctly.
>> That would be nice, but I worry it's too late. The EBML RFC has said for
>> a decade that zero-length integer values represent zero. We're still
>> getting away with rejecting such files in Firefox, per the matroska
>> spec, but not all tools do.
The RFC also says it is a 'draft'. Rightly or wrongly I had presumed the RFC Draft was unofficial and that http://ebml.sourceforge.net/ <http://ebml.sourceforge.net/> was official, but the presentation and relationship of the two specs has been confusing, particularly when trying to determine how to resolve their conflicting definitions.
Discussion at github leans towards saying that a zero-length integer is invalid: https://github.com/Matroska-Org/ebml-specification/pull/15#issuecomment-107623446 <https://github.com/Matroska-Org/ebml-specification/pull/15#issuecomment-107623446>. Also there's a good reason to include zero-length integers I'll send a PR soon for an expanded Data Type section that follows the 1-8 byte requirements of the actual spec, not the rfc draft.
>> http://matroska.org/technical/specs/rfc/index.html sec. 2.4
> Interesting. It's the first time I've seen such a file ever.
Thanks for pointing this sample out.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Matroska-devel