VST3 SDK could need some improvement...

[Markus Krause]

As a dev with 20 years of experience it took me one month to make the VST3 SDK work as expected. Time that could better be invested in updates or new features of the products.

Some of the major bummers I have come across:

- VST 2.4 support does not work in 3.6.10. It is broken
- The shell-script for VST2.4 support on the Mac isn't working
- AudioUnits do not work in 3.6.9. You need to edit the .plist and fix it first. There are also sourcecodes of CoreAudio missing. You need to find them on the web and copy them to the right place first
- Everything is stuffed into a single bloated and messy project-map
- The VST 2.4 example plugin crashes some hosts
- The sourcecodes and resources are scattered all over the place
- A postbuild script for copying the vst files to the right place is missing
- A right-click on the GUI in the wrong place crashes the VSTGUI editor
- CMAKE is failing if there is no VST3 directory present or if the write permissions are missing
- Build files are created within the SDK. This bloats up backups.
...

[matt42]

I feel your VST3 pain

VST 2 is officially dropped, so I doubt there will be any fixes for that. I would suggest posting at the Steinberg developer forum where you'd have a better chance of getting hold of the right people to address your issues

[gnjp]

You could create issues or PRs for these on their Github repo: https://github.com/steinbergmedia/vst3sdk/issues. Some might have already been raised.

[Markus Krause]

All of them are known and have been issued by other developers in the past (Will Perkle et al) . It doesn't seem that there is much interest from Steinberg's side to enhance the situation since a decade. That's why VST3 still isn't established and most developers still provide VST 2.4 versions

[BlueprintInc]

All of them are known and have been issued by other developers in the past (Will Perkle et al) . It doesn't seem that there is much interest from Steinberg's side to enhance the situation since a decade. That's why VST3 still isn't established and most developers still provide VST 2.4 versions

I partly agree, but what does this thread change then or what purpose has it?

[Markus Krause]

I made a poll last month. The majority of the users still prefer VST2.4. After more than a decade VST3 is still not established.
One major reason is the badly designed SDK with many errors and a weak documentation. Obviously Steinberg does not care much about the problems of the developers. Instead of fixing the SDK they try to force the devs to move to the VST3 SDK by disontinuing the licensing of VST2.x.
If we developers don't complain things won't improve

[Angus_FX]

Fair to say that if the smaller plugin development firms weren't such a fiercely independent bunch, we'd have a community-owned alternative by now. None of the big three SDKs are remotely fit for that purpose. Audio Units is closest at this point, but obviously Mac-only (and iOS, if you include AUv3). AAX isn't bad if you ignore the PACE DRM that's baked in to the system, and the highly restrictive licensing model.. well-designed API, crappy corporate politics. It really says something that VST3 manages to be the worst of the bunch, especially following on from the venerable if imperfect VST2.4.

But it's also fair to say, despite all this, those who've tried are not much tempted to try again

[Markus Krause]

I didn't sign the contract for AAX some years ago, because I also didn't like Avid's licensing conditions.
I also refused a contract with Native Instruments for the same reason.

I did a poll last month. Only 1% of the users prefer AAX as a plugin-format. I prefer to live with this neglectable loss-of-income and prefer to waste no further development time on another plugin-format that does more of the same

[ahmadfarihan]

I’m an application programmer but not an audio programmer. What I find it perplexing the audio world is why almost all audio frameworks/dev still depend on Steinberg, Apple and AVID to make it the development of audio apps commercially viable? Is it really hard technically to replace them or is there any IP concern? In all other CS/IT fields there are tonnes of dev and framework options for commercial devs. Am I missing something?

[gnjp]

All these frameworks are propriety, not open source, and tied closely to the development of their respective DAW/OS: Cubase, Logic Pro and Pro Tools (no alternative for AAX). As they are all in competition for your dollar, they have no interest in working together towards a standard. Things like OpenGL have worked well, then you have Apple deprecating it in favour of their own Metal, and continuing to make machine learning more difficult for devs by shunning Nvidia, the list goes on..

[mike_the_ranger]

Any "community" driven standard (you know, the one to rule them all XD) will only be just yet another standard. And I doubt many devs or even users like to support or care about yet another plugin format.

btw personal question to you, Markus: is there anything you like at all? All I see from you are "ranting" topics about all the things you dislike Feels like you don't have fun with anything...

[ahmadfarihan]

All these frameworks are propriety, not open source, and tied closely to the development of their respective DAW/OS: Cubase, Logic Pro and Pro Tools (no alternative for AAX). As they are all in competition for your dollar, they have no interest in working together towards a standard. Things like OpenGL have worked well, then you have Apple deprecating it in favour of their own Metal, and continuing to make machine learning more difficult for devs by shunning Nvidia, the list goes on..

I’m not expecting them to change it, but some other devs/companies maybe with a different paradigm. A new horse in the race, perhaps leading to a slightly different direction.

Apple would do what apple does, but if you are doing machine learning in a server side deployment chances are you won’t use apple anyway. For client side deployment you still need to get all the different machines to test.

OpenGL is horrible anyway.

[ahmadfarihan]

Any "community" driven standard (you know, the one to rule them all XD) will only be just yet another standard. And I doubt many devs or even users like to support or care about yet another plugin format.

btw personal question to you, Markus: is there anything you like at all? All I see from you are "ranting" topics about all the things you dislike Feels like you don't have fun with anything...

Nothing has ever rule everything. And if someone tries to, there’s frodo and the fellowships. It may not even need to be a community driven project, another proprietary library/framework would do. It seems warranted to have a new one at least by now with all the rantings/complains i’ve read on the net.

[Vertion]

What would it take, to make a simple plugin standard for the community?

[BertKoor]

---

What would it take, to make a simple plugin standard for the community?

First and foremost an ecosystem in which developers can do what they do best, and customers have free choice.

In a parallel universum we would be complaining about how badly thought out tthe Jeskola Buzz Plugin API v3.2 is, while we all could have had influence through the mailing list or posting issues on GitHub.