Subject: Re: [linux-audio-dev] latest news: the editor taketh shape
From: Paul Barton-Davis (pbd_AT_Op.Net)
Date: Tue Oct 24 2000 - 04:39:19 EEST
>So you can't write a fade-in effect as a LADSPA plugin, what's the big
>deal!? You can still write n+1 different effects as LADSPA plugins and you
>can use them in Ardour without problems. LADSPA was _not_ designed to
>handle all processing needs and requirements. This design decision was
>made for a reason. However, the LADSPA API does cover many cases. If all
>app developers just say "oh, the API doesn't cover x, I'll use another
>API", nobody is going to use LADSPA. And especially, if a high-profile
>linux audio app like Ardour does not support LADSPA... oh, well.
LADSPA was, by its designers own clear statements of intent, and on
this list recently, constructed as a least-common denominator of
several real-time plugin systems.
If you have a system in which you don't believe you have the CPU power
to run the plugin chain in real-time, LADSPA (or any real-time plugin
API) can't help you.
Ardour is just such a case. I don't believe that most of Ardour's
userbase will have enough CPU power for some time to come to run the
plugins "associated" with an EDL in real time. Ergo, we have to
execute them in non-real-time mode.
As Richard has noted, for many kinds of effects, this makes no
difference, and LADSPA works well. But there are also problems that I
don't think we (or Steinberg, for that matter) ever completely solved
w.r.t plugins that cause the "audio stream" to be extended in time,
such as a delay line. For example, the simple example of an
echo/delay plugin - how does it tell the host that the duration of the
N samples it was handed has been extended to N+M ?
>And btw; I'm still uncertain whether we can distribute VST headers with
>GPL'ed software.
Absolutely. But there are various approaches to this problem, some
rooted in technology, some in back door politics, and several being
looked into right now.
I'm not trying to dump LADSPA. But it wasn't designed with editing
apps in mind, and particular not apps that can't ensure they can run
the plugin chain in real-time. So ideally, I'd like to fix it so that
we *can* use it for such applications.
--p
This archive was generated by hypermail 2b28 : Tue Oct 24 2000 - 05:03:55 EEST