Vadim Zavalishin ( reaktor main developer ) now works at u-he

You can tell a Max device how much latency to report, and it has a useful module that if you wire it up correctly will measure the latency for you.

This assumes the latency is stable. It won't work for dynamically changing latency (as someone asked for in another thread), and the device I built has strange latency behaviour under some limited circumstances. Not serious enough for me to be concerned, I set the latency it has when stable and if it isn't always accurate, the primary use case is for sending midi through another device which adds timing variation, so slight instability is acceptable. It's probably a result of the device being a multi-device which communicates using send/receive, which I've read are known to be a little unpredictable in timing.

Native Instruments could provide something similar for Reaktor, where you measure the latency (with modules they provide) and save with the device, then Reaktor reports it. For most if not all use cases this would be sufficient. You could even measure it for downloaded ensembles and edit them to set it yourself.

But: This feels like a niche problem.

I've never experienced latency problems with any Reaktor ensemble. Admittedly I use mostly the official ones and some blocks based ones from Toybox Audio, and all of these are things which in VST form tend to be zero latency (e.g. I can't find a single VST synth in my collection with anything but zero latency).

To be clear, I am not doubting that some people are making devices which have noticeable latency. I'm only saying that for the average user this is probably not a problem. Would it be good for NI to implement this? Absolutely! Is it a factor in the potential long term survival and usefulness of Reaktor? Probably not.

Hmm i see.

Too bad because Reaktor can be a nice teaching tool for sure in terms of synthesis,sound design. You can obviously go deep with core.

I guess Vadim with his few videos about filter design in Reaktor was a good entry into academia too…i feel this was a good opportunity there to show the technical,dsp side of Reaktor and be taken seriously as this was done at an audio developer conference.

I remember a thread where a teacher was considering using Reaktor to teach his class but he was complaining about something,i don't remember well.

I acknowledged that some devices will have latency. And that this is something Native Instruments should be able to provide using a mechanism similar to that found in Max (though as a software developer I know better than to assume it is simple).

Latency seems to me more a function of whether the processor must delay the signal to process it in the desired manner, rather than complexity. Modern computers are capable of running fairly complex algorithms in real time.

Lookahead requires latency because it must delay the audio signal relative to the control input. Though not all dynamics processors have nor do they necessarily need lookahead. Whether their processing is more "correct" is a matter of opinion requiring a definition of "correct" that may not apply—is the processing of hardware compressors which lack lookahead "incorrect"? It depends on the application. The appropriate questions here are "does it sound good?" and "is it achieving the effect I want?".

If the devices you are building make Reaktor difficult for you without delay compensation, that is understandable. I can see how what I wrote may sound dismissive of the fact that this is a genuine issue for some users, and for that I apologise.

My intent was to address the broader topic, which is whether Reaktor as it is today requires any significant changes to survive, which would warrant considering the departure of a single developer a signal of doom.

I think it's a good decision on Vadim's part. I don't see any progress for REAKTOR development. It has been at a standstill for many years.
The original idea is being lost.

As you can see in my profile, I have switched to hardware; I am now trying to program Haken Audio OSMOSE (Expressive E).

Absolutely! It's not programmed with zeros and ones, but it's a bit like Reaktor if you use the Haken Editor.
The important thing is that OSMOSE uses MPE+ (Midi Polyphonic Expression +). This means that a different midi channel is selected each time a key is pressed, with the possibility of controlling pitch bend (or other) by side pressure, velocity and a very deep (1cm) aftertouch. It also provides midi values when released.
Logic-Pro and some hardware and software instruments can process this midi data.
Via Midi-DIN-In and -Out these can also be sent to Haken Audio CVC (with 440kHz !!!) and used in CV data for my modular system.

Expressive E OSMOSE

Reaktor has a way to grow, and you don't need to claim it has everything it needs...... For you, it may have everything you need, but for me, it doesn't.