Plugin isolation/sandboxing

I didn’t see a feature req post for this, so apologies if one already exists. So Bitwig has the ability to sandbox plugins, either by individual or by mfg. This increases memory usage a bit, but it also allows for graceful crashes in case a plugin goes haywire (i.e. it doesn’t take down the whole program). It helps make for an absolutely rock solid DAW. I’m still getting all the kinks worked out of my Win10 GP setup, but the last 2 nights, one plugin of mine (Z3ta 2) just randomly decided to start crashing GP whenever it would get loaded. I was using predictive loading and even though it was 3 racks down - it would crash the entire thing. Even though it worked fine for almost a month, It decided to start happening on friday night (and again last night) randomly. Fortunately the event log pointed me to that particular plugin - which I promply renamed and had to rescan the entire plugin list (in the middle of Don’t Stop Believin’ mind you) while the band played without any piano.

This same sort of thing happened to me in the beginning with OP-X Pro II - I ended up just getting rid of it and replacing the Oberheim sounds with the free OBXd (which is rock solid - albeit not quite as good). The developer basically told me that it’s a GP issue and that older versions of GP worked fine with it, but that’s a different issue. Plugin sandboxing would have saved me a ton of headache at the gig.

Plugin sandboxing is a good idea in a DAW where latency does not matter.
But in live situations the latency would be too much.
And in a case of an issue gig performer would not crash, but you would not get any sound from the crashed plugin.

This are the reasons why plugin sandboxing will not be really useful for live situations,
Better your plugins are reliable.

1 Like

I assume you are talking about loading latency. Isn’t that supposed to be what predictive loading is supposed to solve for?

TBH, I’m fine with that. On a rack where I have 5-7 splits/layers across 2 keybeds, one sound not playing is not the end of the world. The entire host crashing takes me out for an entire song (if not longer) and sucks when I have to try and debug which plugin is taking the whole thing down while the band is giving me dirty looks.

this one burned in for nearly a month with no issue and decided to crop up just on a gig out of nowhere. These are also commercial, paid for plugins, not freebies. I’ve used both Z3ta 2 and OP-X Pro II for nearly 7 years on my home DAW machine (Cakewalk, Bitwig, Reaper) and NEVER had crash problems with either until I started using them on GP. How long do you let a plugin burn in before you decide it is stable enough to use on a gig? 3 months, a year?

I’m not looking to start an argument, but this is just my perspective based upon someone who has been gigging as a keyboardist in A LOT of bands for a long time and have come into this realm after working with almost exclusively hardware live for more than 10 years now. Sometimes I have a week to build an entire setlist for a band - that includes learning the tune, making notes, and programming everything. I’ve had hardware completely take a dive (my Kronos has crapped itself several times) onstage on high profile gigs and I have had other “reliable” synths like my old Yamaha Motif decide to randomly play out of tune years ago. Nothing is perfect. You deal with it and move on, trying to improve your situation and come up with better failsafes. The whole reason I even went to GigPerformer is because the Gemini synth module in my organ controller decided to short itself out (through a $300 Radial DI) with some goofy current coming from our mixer in the middle of a gig, leaving me forced to play 3 hours of a gig with nothing but a Behringer string synth. Once I get the module back, I’ll have a backup, but my feature request stands as it is, even though you may not think it is necessary.

No I am talking about latency when playing live.
Could you live with a playing latency of at least 25 ms with plugin sandbox?

When you got a crash, did you contact the plugin manufacturer?
I did in the past with spectrasonics, arturia, synaptiq to name a few.
They all reacted on that issues ans provided a fix or are at least developing a fix.
The only thing they really need is a reproducable testcase.

I did with OPX Pro II when it happened weeks ago. The developer blamed it on GP because, according to him, it worked perfectly with earlier versions. Regarding Z3ta - no, because this just happened with that this weekend and I’ve been busy gigging.

I don’t have those sort of latency issues, even when doing full isolation in Bitwig. Even on my older DAW cpu, which was slower than what I am using with GP now.

Which latency can you play plugins live in Bitwig?

This has actually been on our list for a long time but I can’t say when it might happen. This is one of those things that is much more complicated when you allow a general graph (i.e. anything can be connected to anything else) rather than a simple channel strip.

Why did you have to rescan the entire plugin list?

Can you reproduce this issue? In the early days (2 generations ago) of GP, it was definitely the case that GP struggled with certain plugins and we spent quite a bit of time handling outlier issues presented by some plugins. These days, it is very rare to find a plugin whose crash is due to GP but you never know. We have worked successfully with plugin manufacturers to resolve such issues when they occur but we need a reproducible case after which point we can easily determine the problem and what is at fault.

Generally, something has to have changed somewhere for things that worked fine for a long time and then caused problems. Any Windows updates, antivirus updates, installation of other programs that could have replaced system DLLs that were previously being used by that plugin?

(NB I’m not objecting to your request for plugin isolation, just trying to understand why something that was working perfectly would suddenly not! There’s no magic involved, something had to have changed somewhere.)

TBH - I was on stage under the gun (it literally crashed about 4 seconds before I was supposed to start playing a song with a piano intro that the band cued off of) and I wanted to remove all traces of it and I hit the wrong button. I’ve disabled plugins in the past only to notice that on racks that still have them still play them, so I was mad frantic, just trying to keep the stupid thing from crashing again. Fortunately, I had zero issues with it after that (I went back at the end of that set and tried to just quickly replace the sounds presets from other synths that were moderately close, just to get me thru the gig). The experience I had last night brought back the same feelings that I felt years ago when I decided to abandon computers for live use and go back exclusively to hardware. That’s not my intention here though - as I know all of these issues are solvable (hence my post and general participation on this forum).

I don’t see why not. All my gear is still packed up, but I should be able to get things set back up tonight. Also, I should still have the crash entries in the windows event log (which I assume you would want) that I could pull up as well. How long are the crash logs held on disk before they roll over? I can definitely provide info for both plugins.

With all of the issues with Cakewalk and Gibson, I’m not sure if the “new” Cakewalk/Bandlab is supporting Z3ta2 anymore, but I’ll definitely reach out to them as well. They at least still have it available for download, so there may be something there.

that totally makes sense, and I appreciate the complexities involved with all this stuff… Especially for a product that nearly 100% of its primary functionality relies on the stability of 3rd party libraries that you have little or no control over :smile:

Meanwhile, I travelled to Norway to do a show last month and while GP worked great, one of the keyboards stopped working…trust me, that’s an even bigger problem :disappointed_relieved:

I actually carry a spare laptop that completely replicates my software environment specifically for issues like what you describe although I have to say that in the last four years of touring (we had GP internally for about a year before we released it) I have never had to switch laptops in he middle of a show. But I never change ANYTHING on those laptops unless I have at least a month to do serious testing.

If the crash entries include stack traces so we can see the sequence of calls that led to the crash then such info might be very useful. But we really need a reproducible case.

haha yea. See my earlier post - I know the grass is never greener. I joked with the band last night that they need to add to our rider with the promoter that the venue needs to start providing me a tuned, maintained and working acoustic piano and B3 at every gig. I’m still waiting to hear back on that. :smile: It’s not that bad in the grand scheme of things though… if it was, I’d just go back to playing drums.

shouldn’t be an issue - I’ll get everything loaded up tonight after the kids go to bed and see if I can break it. What do you need? Just the crash reports?

Ideally a reproducible way to break it.