23 Jan
2008
23 Jan
'08
7:44 a.m.
Steve Harris wrote:
On 23 Jan 2008, at 10:10, Krzysztof Foltman wrote:
An interesting experiment might be to write a VST and/or AU wrapper
around LV2.
The interesting things about this would be:
* LV2 support in apps like reaper is (at least partially) present
* a lot of open source LV2 plugins get a larger userbase (which might
contribute to the plugin's quality, and maybe also to the adoption of LV2).
* It could also serve as an example for people that are used to writing
VST or AU plugins on how to write LV2 plugins. Some sort of
'documentation by equivalence' or so.
* It might reveal some design issues with LV2 (if there are any :))
The downside would be that there is less motivation to implement both
LV2 and VST since such a wrapper gives you LV2 through VST. But then
again, LV2 can do so much more than VST, no?
Just a thought...
Pieter
PS: I haven't looked at LV2 nor VST (yet), so technically I should
probably not participate. But nevertheless I do :).
Steve Harris wrote:
Hah, you're right, though there are some people who have a knack for,
and get a kick out of writing conformance tools.
Like Nick Lamb, who wrote Demolition for LADSPA:
http://devel.tlrmx.org/audio/demolition/
I don't think anyone ever wrote a set up conformance plugins for
LADSPA that measured host conformance though - that would also be
useful. To my mind it's better for us to develop a
large suite of tools
and plugins to demonstrate the viability and advantages before we go
I think we
indeed need lots of testing tools - like debugging hosts/
plugins spiked with lots of pre/postcondition checks, or even some
validity checking libraries that could be easily inserted
(#ifdef'ed) into "real" hosts/plugins, to check plugin/host
behaviour in "real world".
I recall what incomplete/buggy standard implementations did in Buzz
and VST worlds, and it'd be nice to have some tools to prevent
repeating the same nightmare.
As for "who will write it", I guess it can't be a single-person
project, because just one person is unlikely to come up with *all*
the useful checks (and it would be incredibly boring anyway).