[LV2] Killing the event extension

David Robillard d at drobilla.net
Wed Jan 29 07:49:09 PST 2014

On Wed, 2014-01-29 at 06:56 +0100, hermann meyer wrote:
> Am 29.01.2014 00:57, schrieb David Robillard:
> > On 14-01-28 04:38 PM, hermann meyer wrote:
> >> I would love if we talk about hosts implement the basic features, like
> >> the work extension, but I'm not aware that there is a concurrent way of
> >> implement the event extension.
> >
> > I don't understand what you are getting at. What does the work 
> > extension have to do with anything?
> Simply,
> I, would love if you would start to "presure" host author’s to include 
> the work extension instead talk about that you would start to "pressure" 
> host author's to remove the event extension. ;-)

I am.  There is not much more I can do at the spec level to do so: it's
stable, it's a recommended extension, one of the official example
plugins uses it, it's supported in all the hosts I personally work on.

As far as LV2 itself goes, there is no problem with the worker
extension.  It's there, its purpose is clear, there is (and only ever
has been) one extension that does that, it's supported in some things,

If you want a host author to implement the worker extension, then file a
feature request.

>From your perspective, sure, you want host authors to implement
everything, including backwards compatibility nuisances, ponies, and
helicopters.  Host authors don't want to do anything but call run() and
want plugins to deal with everything.  I don't have the luxury of
completely ignoring either.  The fact of the matter is that the
event/atom problem is a barrier to LV2 adoption and a wart on the spec,
and the worker extension is not.  Do I want the work extension
implemented?  Sure, I think it's a great facility.  Is it actively
damaging the LV2 project itself?  No.

> > There is, at this point, no point in plugins implementing both (which 
> > indeed, you can't easily do). That's the point of this thread. Just 
> > switch.
> >
> >> However, now big deal. As I said, I'll just remove the midi extension of
> >> gxtuner.lv2 and be done.
> >
> > Why? Removing it is probably just as much work as updating it. That's 
> > silly.
> >
> Because I never made it to the atom message struct. I've spent some 
> weeks with it, before I've decided to use the event extension instead. 
> Removing the midi output of the tuner is a easy task.

Well... deliberately using the deprecated extension *after* trying the
new one was a bad decision.  Sorry.

If you're just using MIDI there is not much difference between either,
the sizes and structs used are just a bit different.  If you had a
problem, you should have asked, and it would have been
resolved/clarified.  Lack of communication like this is what makes it
necessary to put pressure on projects via means like warnings and
removing support.  I'm not omniscient, I can only use the information
and tools that are available to move this project forward.


More information about the Devel mailing list