Will NI be releasing TestStand for Linix in the future?

[John_Speckhard]

Will NI be releasing TestStand for Linix in the future?

[Dennis_Knutson]

Since everything is based on ActiveX and the Linux market is incredibly small, my guess is that it will never happen. An official response from NI May say something different.

[HeroOfHyrule]

TestStand does not currently support Linux. I am not sure what our future plans are for this, but we like to be aware of what customers are interested in, so I will make note of your interest in a Linux version of TestStand. Thank you for the interest.

[Norbert_B]

Point is, that Linux has a much wider variety than Windows referring kernel modules and "flavors".

So when considering "Does a company provide applications for Linux", it always has to select which distributions are (officially) supported, which are not.

So: What is your Linux distribution?

All in all, ActiveX is limited to the Microsoft world, so Dennis is absolutely correct. Porting an application basing on a specific framework component part of a specific OS (type) to something more "open" is a huge effort in development and therefore not cheap. If the market is big enough to justify that effort, i think NI WILL port TestStand.

But i wouldn't expect the market to be *big enough* for this right now.

Norbert

[JonP]

My company would also be interested if some form of TestStand was supported on Linux. Perhaps the ActiveX interfaces could be implemented with something different underneath, like RPC?

Obviously only the Linux distributions NI handles for LabVIEW would need to be supported.

[Dennis_Knutson]

You should post this in the Idea Exchange if you want any visibility.

[sagray3]

I think that the cost and space savings of running TS & LV on a Linux platform vs a Win10+ platform are obvious. 

Not to mention the reduction in the number of security and platform-management issues. 

The first question is "can it be done". 

If the only thing preventing it is integration with OS, and MS is largely in charge of OS development now that it's in charge of GIT and has tools out already for mixed-mode & targeted development, then I would think that such issues won't take long to resolve. It's really a matter of which distro, which group of programmers are racing to make this happen. 

Will it dramatically affect the cost of a rack full of test-equipment? Probably not. Not hardware-wise. 

But there are other factors to consider. Someone has to sustain that rack & the test program running on it. 

As of right now, TS & LV are locked to Win10 running on an X86 platform. 

And that might not even see long-term sustainment from MS without some sort of fee platform, just as with Win 7 & Win XP.

Second even if it doesn't seem cheaper in terms of HW up front, the idea of a Linux embedded platform running NI software means a drop-in upgradable replacement processing module with the test code already embedded in firmware. That will be the size of a cellphone battery inside of 3 years.

There are too many forces moving to remove a PC from the equation to ignore Linux as the OS for shrinking, rugged and secure TS platforms. 

[Beutlich]

Are there any efforts for ActiveX porting other than through OS emulation/virtualization like WINE? That sounds like an even bigger effort than refactoring the TestStand Engine to non-ActiveX dependencies, but it might have a wider community participation.

TestStand on Linux sounds interesting, but TestStand still expects to be the Operator Interface, and the graphics features in Windows make it attractive over Linux for HMI, especially for something as complex as a sequencer. It is easy enough to make a controller in Linux for a system, and then integrate the TestStand machine with that controller. I like this model myself because I already have separate controllers for different subsystems (which decouples control for operations and maintenance), so tacking on a specific TestStand machine does not incur an additional cost adding a network architecture. You could practically do the same thing across an emulation/virtualization if you had only one computing device.

I've wondered how the Windows IoT OS would hold up for merging the easy graphics of Windows with reliability demands of embedded systems. I've seen some Windows IoT devices used for HMI, but I haven't tried TestStand on a unit and I don't know anyone seriously considering it.

[sagray3]

hm, seems that the real problem here is TS, not Linux. 

TS as it stands is more of a gilded mousetrap than a real solution with inherent sustainability. 

It's the same problem that developers had building 32-bit Win32 C apps in VS6 to run on XP. 

The world wasn't going to stick with C, Win32, VS6 and XP forever. 

The forward-looking developer had to contend with this. 

The code that they wrote one day either had to be ported forward somehow, or the platform had to be locked-down and sustained. 

Plain and simple, developing in LV and TS creates some major roadblocks for code portability and future-proofing. 

This is why Linux has even entered the conversation. 

And that is why many developers do their utter best to avoid developing in Windows in the first place. 

Which raises the possibility that worrying about migrating TS code to Linux is a lot like trying to make your truck & trailer combination faster by putting more horses in the trailer. You don't develop in TS, as it sits, if you really want to run Linux. It's more likely that you do something like write some drivers in Python and then call them in a way that is similar to running javascript apps that make use of HTML5 applets for the graphical parts. That's not as robust or as easy as writing tests in Labview but obviously there's a trade-off. Is it more work? Of course. But you're paying programmers to write code, right? The more code that is written in LV and TS, the more that your entire test is dependent on Windows and if you really wanted that, then you wouldn't complain about the lack of TS support in Linux. If you want "open, platform-neutral development" with a future-proof OS then you wouldn't touch TS, LV & Windows with a 10-foot pole and the OS code that you develop using neutral scripting languages will run on Linux, Windows, MacOS, whatever, all without upgrade issues. As long as your drivers and js didn't get broken, and if so they're still probably relatively easy to fix. No way am I saying that this would be easy for anyone but competent Open Source developers. But there are a lot of competent OS developers out there. 

So my guess is that if NI is thinking of "porting" TS to Linux, they're thinking more along these lines. 

ActiveX is 1990 technology. It took MS time to develop it and it took developers time to build on it. 

But it's 30 years out of date, today.

A Steam developer wouldn't even think of using it, and this is the main reason why.

[WireWeaver]

I don't have any big updates to share on the roadmap for TestStand with regards to Linux support but it is certainly on our radar and is a topic that I'm interested in discussing. The attitude and ecosystem around Linux has changed and NI has been investing to be supportive (New approach to Linux Drivers, NI LinuxRT, etc..).

I also want to reiterate what Norbert mentioned already:

---

If the market is big enough to justify that effort, i think NI WILL port TestStand.

Posting to the idea exchange (already mentioned above) is a good step to express this need. I'm surprised that nobody has done that given the interest in this thread. It'd be helpful to comment on what kind of support you need (what distro?, OI only or editor capabilities?, remote OI or editor?, etc..).

Thanks,

Trent Weaver

Product Manager | Test Software

National Instruments