Subject: RE: [linux-audio-dev] extending LADSPA,
From: Richard W.E. Furse (richard_AT_muse.demon.co.uk)
Date: Fri Oct 27 2000 - 17:19:18 EEST
I like this broadly - some thoughts from a LADSPA perspective:
STREAMING
Logically very similar to LADSPA (simple block in/out) - most of the faff
of LADSPA is nailing down how the plugin describes itself to the host to
allow this to happen! The only addition I spot immediately is the provision
of historical input. This is missing from LADSPA and would require extra
logic in all hosts - plus the plugin has seen that data anyway and can
remember it if it wants to, so this isn't a necessary feature. The 'S' of
LADSPA strikes again...
I like the sound of gain maps etc - hopefully LADSPA handles such issues
through control signals, as the approach here perhaps could. Beyond this,
I'm very reluctant to build such nice features into a plugin API as it
requires additional sophistication in all hosts.
EDITING
I don't know how your are/if you want to support control signals (envelopes
etc) for this. Beyond this the only problem that sticks out is the
requirement to provide complete input signals to the plugin in RAM. This is
fine in a RAM-based sound editor, but is rather unfortunate while trying to
edit a Wagner Opera. (It also doesn't allow nice stuff like real-time
control of granular sound synthesis from a static RAM-based grain source,
but this is one of the more obscure features of my somewhat rambling API
suggestions yesterday!)
I suspect your host is highly compatible with LADSPA plugins: existing
LADSPA plugins can be wrapped directly for streaming use and block use.
Plugins written using yesterdays suggestions would also be straightforward,
as the heart of your implementation of move_noncausal_inputs() would be
something like connect_port(stream.start + Offset).
Good luck - and thanks for an interesting first posting.
--Richard
-----Original Message-----
From: Paul Sladen [SMTP:paul_AT_sladen.org]
Sent: Friday, October 27, 2000 12:49 PM
To: LAD
Subject: RE: [linux-audio-dev] extending LADSPA,
Ok, I've not studied the LADSPA API, but partly on purpose, because of
what I'm working on; but this seems to answer and enable everything that
people have asked for.
The basic idea is that the plugin is passed not only a memory location and
length, but /also/ it's location within a larger buffer.
[...]
This archive was generated by hypermail 2b28 : Fri Oct 27 2000 - 17:53:33 EEST