[LAD] LADSPA extension for periodic control values?

[Stefano D'Angelo]

2009/6/19 Stefan Kost <ensonic at hora-obscura.de>:
> Stefano D'Angelo schrieb:
>> 2009/6/13 Jörn Nettingsmeier <nettings at folkwang-hochschule.de>:
>>> hi everyone!
>>>
>>>
>>> sorry if this has been discussed before, but i didn't find anything in
>>> the archives...
>>> consider the case of periodic control values of LADSPA plugins, for
>>> instance the azimuth in a horizontal panner or the phase shift in a phaser.
>>> currently, they are usually marked as BOUNDED_BELOW and BOUNDED_ABOVE,
>>> but the host has no way of knowing that the upper bound is next to the
>>> lower bound, so that it can chose the shortest path to the next value
>>> when interpolating automation control points.
>>>
>>> take ardour, for example: if i want to spin a source 360 degrees, i have
>>> to start at 0, set a control point at 180, set another control point at
>>> the exact next sample to -180 and then onwards. if there is even a
>>> single sample between the control points, the interpolation will cause
>>> the image to jump in weird ways, because it doesn't know that 180 == -180.
>>>
>>> does it make sense to add a new hint to LADSPA, something like
>>> LADSPA_HINT_PERIODIC? it would mandate LADSPA_HINT_BOUNDED_BELOW and
>>> LADSPA_HINT_BOUNDED_ABOVE as well as the respective port range hints,
>>> *and* imply that LowerBound is equivalent to UpperBound in the port
>>> range hint structure.
>>>
>>> this would enable hosts to do the Right Thing(tm).
>>
>> Hi,
>>
>> I think it makes no sense nowadays to do something like that, for two reasons:
>>
>> * LADSPA is old, hence old LADSPA hosts will not support that and
>> new/maintained ones are not likely to support that, since many are
>> focusing on newer APIs;
>> * An LV2 extension would allow you to do that quite easily in LV2,
>> which is probably preferable these days.
>>
>> That said, I can already hear people who disagree with me shouting and
>> screaming.
>
> I disagree. I would not mind adding support for more hints to the gstreamer
> ladspa wrapper. Its not a lot of work and its good to have that now even if lv2
> takes over in the long run.
>
> The other think I'd like to have is hints for audio-port channel
> grouping/mapping. That is describe that a port is the left-channel of stereo
> pair named <xyz>, Even if this goes to the rdf.
>
> Stefan

Stefan, once again, I originally misunderstood the post by Jorn
thinking he wanted to add that stuff outside of the LADSPA header, the
discussion went and is still going on.

Stefano