[linux-audio-dev] LADSPA 2 decision points

We haven't heard from a number of key people yet, but I think the overall
impression is positive so far. There are a number of decisions that need
to be made if it's going to go ahead. Either things that I didn't
consider, or where I made arbitrary choices:

A) Dropping runAdding: personally I dont think runAdding is worthwhile. How
many hosts take advantage of it? Its something LADSPA 1.1 can do and 2.0
wouldn't be able to though.

B) RDF syntax: RDF/Turtle seems a lot more popular in these parts than
RDF/XML. We could mandate Turtle for all LADSPA metadata.

C) Port scales: these weren't really in LADSPA 1.x, but they were in lrdf, so
adding them is a bit arguable. I think they're clearly beneficial though,
and they are getting close to critcal mass recently.

D) Port shortnames: for OSC, Pd etc. access. I dont have a strong feeling on
that. It /is/ an extension, but retrofitting it will mean OSC servers etc.
will have to support both accessor methods.

E) Bundles: no-one screamed when I suggested it, but its a bit different to
LADSPA 1's /lib directory. Hosts would be required to allow the plugin
link to libraries in the bundle if they need to (this is why OPENSTEP did
it, I guess its done by setting LD_LIBRARY_PATH, but I dont know), which
is a bit more work, but makes moving plugins between machines easier. Oh,
and the plugins needs to have the path of its bundle passed to it when its
instantiated, which isn't in the API right now.

F) Host OS specific macros in ladspa-2.h that Jack O'Q suggested. It seems
like a reasonable idea to me. No need to punish users of inferiour OSs
more than they are allready ;)

- Steve
Received on Sun Apr 23 00:15:12 2006