[Bug 751735] dashdemux: incorrect parsing and handling of segment templates

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=751735

Sebastian Dröge (slomo) <slomo at coaxion.net> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |slomo at coaxion.net

--- Comment #3 from Sebastian Dröge (slomo) <slomo at coaxion.net> ---
(In reply to Thiago Sousa Santos from comment #2)
> I'm not sure about this one, I think your interpretation is correct because
> the spec tells to replace the $Number$ with the value and then the 2nd
> $Number$ would lose its first $ and would not be replaced, but what if we
> start replacing from the end towards the start or if all replacements should
> be done 'at the same time'? What interpretation could have us covering most
> of the valid use cases?

I would make it left to right, and if stuff is not consistent make it an error
or just fail with the resulting 404. As the spec uses $ as the opening and
closing "parenthesis", it's not unambiguous and there can't be a solution that
catches everything.

> > {"TestMedia$Time", NULL},   /* Identifier not finished with $ */
> > {"Time$Time%0d$", "Time100"},       /* usage of format smaller than number
> > of digits */
> > {"Time$Time%01d$", "Time100"},      /* usage of format smaller than number
> > of digits */
> > {"Time$Time%05d$", "Time00100"},    /* usage of format bigger than number of
> > digits */
> 
> > {"Time$Time%05dtest$", "Time00100test"},    /* usage extra text in format */
> 
> Is this valid? I couldn't find anything in the spec that allowed suffixing
> that was not the %0d identifier.

IIRC it's valid, but even if not I wouldn't surprise if streams with that
appear out there. It seems too intuitive that this should be possible.

I don't think it's great that C format strings were used here because of that
reason.

> > {"Time$Time%05u$", "Time00100"},    /* %u format */
> > {"Time$Time%05x$", "Time00064"},    /* %x format */
> > {"Time$Time%05utest$", "Time00100test"},    /* %u format followed by text */
> > {"Time$Time%05xtest$", "Time00064test"},    /* %x format followed by text */
> Are %u and %x allowed by the spec? It seems that it should always be a %d
> but as it is a bit dubious I'd be ok for accepting those as well.

The spec only allows %d, yes. But because it's a bit dubious as you say, we
also support %x and %u. In mpd parser.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.