On Tue, 2006-04-25 at 23:11 +0100, Steve Harris wrote: > On Tue, Apr 25, 2006 at 06:01:18PM -0400, Dave Robillard wrote: > > On Tue, 2006-04-25 at 22:47 +0100, Steve Harris wrote: > > > I'm in a cruft killing mood. > > > > > > Has anyone ever used ImplementationData? I know I haven't, and I goggled > > > for it, but all I found was the ladspa.h file, which has a comment to the > > > affect of: we're not sure why you'd need this, but, just incase here's a > > > void *. > > > > > > If anoyone has used it, or has a potential use for it, it should stay, but > > > otherwise it should go IMHO. > > > > I was just thinking this while going through the header (cleaned it up a > > bit). I don't even really understand what it's purpose is to be honest, > > never have. > > > > What does it allow that would not be possible if it weren't there? > > Well, it somewhere to stash "class" level information, but I just use > statics (easier) and/or shm (harder, but cross host) for that. > > It would be a bit tricky to use as you'd have to be really sure that the > host was only going to request one descriptor, otherwise you'd have > multiple copies of your "class" data, which could be either wasteful or > dangerous. Ah. It makes sense I suppose, but class data can be static or just put in the LADSPA_Handle anyway.
I say if it /can/ go without loss of functionality, it should go, so let's check... I just grepped my plugins source directory, which I'm relatively certain contains every LADSPA plugin ever released (and some that weren't). Other than some (unused) boilerplate in CMT's C++ wrappper, and setting it to NULL, it's not used once. -DR-
