If JUCE requires a huge IDE to compile programs, it sounds like an overly demanding framework...
You can build JUCE from the command line if you so wish. It doesn't need an IDE, thiink of it as a collection of C++ classes
I try to keep tools & dependencies to a minimum
Agreed. Keeping dependencies to a minimum is just about the
only way you will ever create something which is actually deployable on a linux system other than the one you developed it on.
Also it has some bugs that probably won't be solved for a long time..
@sadko4u: I think your approach is wrong. You are trying to force GTK into the JUCE framework, and, whether you perceive that as a 'bug' in JUCE or not, you are going to have problems further down the line, even if you get over that particular issue. Likely you will end up with a 'VST' which only works in a GTK host application (so that kind of excludes any host except Ardour) and more specifically a GTK 'n' application so that means problems with any GTK host which isn't using the
exact version of GTK you are using. (It's now just about enshrined in the GTK spec that incompatibility between versions is a design 'feature' which may not help the situation).
For my own plug-ins I developed my own X11 based graphics 'engine'. At the present time this is what you have to do to integrate with a linuxVST host. Fortunately JUCE provides this too, but, you have to use JUCE's UI framework - that's the point of it. So, fundamentally you will have to rewrite your UIs to either use JUCE, or X11 directly. That probably seems like a lot of work (it turns out writing plug-ins is actually quite difficult, especially for linux), but its still probably less work than you will have to do to make GTK work reliably.
LV2 is not the 'magic bullet'. - I've developed for just about all plug-in formats on all the main OS at one time or another and the only thing that holds true is that no plug-in API is ideal, and likely there will never be one which is. Many developers don't like VST for linux because of the licensing, but there are indications that is changing. I suspect LV2 will always have a problem gaining acceptance outside of Linux. Fundamentally it exists to (try) and solve some very linux specific issues. Outside of that there is no
essential reason for it on any other platform. For it (or any other new) format to gain acceptance, you would have to write a 'must have' plug-in, and, then decide to only release it in a format that just about no-one supports. For reference look how long it has taken for Steinberg's
own VST3 standard to gain acceptance.
This isn't anti-LV2 or pro VST, its just based on my experience.