Subject: Re: [linux-audio-dev] misc ladspa questions
From: Paul Barton-Davis (pbd_AT_Op.Net)
Date: Mon Mar 20 2000 - 20:20:05 EET
>1. Those plugins themselves are indifferent to their input/output ranges -
>they will work in either scheme. It is the host that uses 16bit implicitly
>as it only reads and writes 16bit to file.
>
>Personally I don't see a point to the [-1,1] restriction - most plugins
I feel more strongly than this. I vehemently oppose such a
restriction. It removes a lot of the point of 32 bit floating point,
and *critically* it prevents bit-for-bit reproduction of single 24 bit
signals. If I read a 24 bit integer from a digital interface, I
*absolutely* expect to get exactly that value back when its output, if
no processing is done to the signal. a [-1,1] range makes this all but
guaranteed to fail.
>2. I think hosts should be free to use their own plugin-location
>approaches, but a standard location for shared plugins would be rather nice
>- really good idea. I'm no expert on standard UNIX file paths, but one day
>Linux may ship with a bunch of plugins - so perhaps a non /local/ directory
>would be appropriate? Should we be using environment variables to determine
>path?
Yes. My preferred order for this:
1) see if there is an environment variable
2) read a config file to (potentially) override it
3) read the cmd line to (potentially) override it
We should just define the name of the environment variable, and then
each host can do its own thing from there on. I would suggest, duh:
LADSPA_PLUGIN_PATH (its a colon separated list of directories to
search).
--p
This archive was generated by hypermail 2b28 : Mon Mar 20 2000 - 21:05:10 EET