[vorbis] Vorbis determined to be as good as MPC at 128 kbps!

[Jack Moffitt]

> And just look at Garf's tunnings.  Perhaps not perfect, but better than the
> ref design, but Monty was 100% unwilling to put those improvements into the
> official version, or even let Garf help merge it into the official one.
> 
> The Garf tunings weren't perfect, no.  But Monty was unwilling to even
> consider joint tuning / development so that higher bit rates could be
> improved.
> 
> In fact, Monty's resistance and refusal is the stated sole reason that Garf
> ended up giving up and totally abandoning the Vorbis format.  It takes a lot
> of frustration for an early major supporter like that to just give up.

We've not always been the best at welcoming others to the table.  Vorbis
is a hard project, and most of the people who show up have good ideas,
but not enough experience to pull them off.  We do try and help people
along, but in the case of things like tunings or core encoder changes,
these have to be carefully reasoned and thought about, which takes a
significant amount of time.  As with anythign that takes so much effort,
those things tend to get put to the end of the list, or procrastinated
on.

In any case, we're trying to improve in this area.  Garf never said
anything to me about his frustration, but then again, there have been
periods where I haven't monitored the lists or IRC.  I hope he's not
gone for good.

> It's been commented officially too many times that Xiph expects other people
> to develop their own versions & tunnings, and that those 3rd party
> improvements will not be put into the official encoder.

Please provide references of these official notices.  We expect others
to do some work here and there, but I think the plan (as with any open
source project) is always to fold back in changes that make sense.

Optimizations beyond a certain level don't belong in a reference
encoder.  That's about the only thing I can think of where we've refused
to accept major work, and I think there's a good reason to do this.  I
don't think we've (ie, Monty or I) have ever said anything with regards
to tunings.

> It's more than a little unrealistic to expect that Monty is going to manage
> to develop his own tunings that just happen to be the best, and do so
> without any help from any other people.

Everyone is agreed on this point.  Monty is mortal and human just like
the rest of us.

> So, unless Xiph & Monty actively embraces other developers and sets up a
> process where their changes can be folded into the official version, it's
> not going to improve much compared to the third party ones.

I will try and rectify this problem.  Please let me know if you have any
specific suggestions.

> Not every change they make will be better, no.  But there needs to be a
> willingness by Xiph to fold the real improvements into the reference
> version, and a process by which developers can submit potential
> improvements.  Xiph has neither.

We have the same process every open source project (or most every) has.
Send patches via email, which will be discussed, etc.  Many patches have
made it into Vorbis from a wide variety of people in just this fashion.
What we don't have is any kind of 'keep track of progress' or real
feedback mechanism for non-code changes.  Tunings fall into this
category.  How should this work?

> And that's why so many people tend to call these other versions "forks"
> because they tend to feel that those improvements will never ever be put
> into the official code.

I'll put on my official hate and say "We do want third party
contributions to be folded in where appropriate."  Goods tunings I think
are such an instance.  Let's try to do what we can.

> Considering the speed of past development, and that focus is now on Theora
> and Vorbis-II, I think the odds are extremely remote that the Xiph version
> will ever match, much less surpass, the already existing third party tuned
> versions.  Xiph's attitudes towards Voribs has always been "We made the
> specs and provided sample code.  Improving it is your job."

We have multiple teams working on many different things.  Monty
is not focused on Vorbis II, but on the contrary is hacking on OggFile
so that multicodec streams will be as eays as single ones.  Ralph and
some others are focused on Theora codec work.  We need to ship Theora,
and we need to give others the tools to deal with Theora files.

We try to do as much as we can.  We're only a finite amount of people,
and only one of us works on this full time.  I'm sorry we can't drop
everythign else to audition every patch from every developer as soon as
it comes in.

That said, I know we aren't the best at it, as I've said.  And to our
credit, it's a hard problem.  But we want to get better.  And we will
get better.  Please, just have a little patience :)

> Maybe things have changed.  If so, Xiph and the Voribs team have a lot of
> developer relations damage to undo.

I'd like to think we haven't 'damaged' anything.  But anyway, we want to
improve in this area.  Suggestions?

jack.
--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.