Instrument Control (GPIB, Serial, VISA, IVI)

cancel
Showing results for 
Search instead for 
Did you mean: 

Is this a VISA Bug/User Error/Other? (NI-VISA16.0.0 on Scientific Linux 6.8)

Solved!
Go to solution

After successfully installing Labview 2016 and NI-VISA on a new SL machine, I attempt to run Visa Configuration and get the following:

 

$ visaconf
Can't load library /usr/local/lib/liblvrt.so.15.0
libXinerama.so.1: cannot open shared object file: No such file or directory
To download the LabVIEW Runtime engine, go to http://www.ni.com/rteFinder?dest=lvrte&version=15.0&platform=Linux&lang=en

But when I attempt to install the 2015 RTE I get pushback from the 2016 RTE:

./INSTALL

Preparing...                ########################################### [100%]
	package nissli-16.0.0-f0.i386 (which is newer than nissli-15.0.0-f1.i386) is already installed
	package nissli-16.0.0-f0.x86_64 (which is newer than nissli-15.0.0-f1.i386) is already installed
	package nisslcerts-16.0.0-3.0.noarch (which is newer than nisslcerts-15.0.0-3.1.noarch) is already installed
	package nicurli-16.0.0-f0.i386 (which is newer than nicurli-15.0.0-f1.i386) is already installed
	package nicurli-16.0.0-f0.x86_64 (which is newer than nicurli-15.0.0-f1.i386) is already installed
	package nicurli-16.0.0-f0.i386 (which is newer than nicurli-15.0.0-f1.x86_64) is already installed
	package nicurli-16.0.0-f0.x86_64 (which is newer than nicurli-15.0.0-f1.x86_64) is already installed
	package nitdmsi-16.0.0-f0.i386 (which is newer than nitdmsi-15.1.0-f0.x86_64) is already installed
	package nitdmsi-16.0.0-f0.x86_64 (which is newer than nitdmsi-15.1.0-f0.x86_64) is already installed
	package nitdmsi-16.0.0-f0.i386 (which is newer than nitdmsi-15.1.0-f0.i386) is already installed
	package nitdmsi-16.0.0-f0.x86_64 (which is newer than nitdmsi-15.1.0-f0.i386) is already installed
	package nisvcloc-16.0.0-1.noarch (which is newer than nisvcloc-15.0.0-1.noarch) is already installed
	package nissli-16.0.0-f0.i386 (which is newer than nissli-15.0.0-f1.x86_64) is already installed
	package nissli-16.0.0-f0.x86_64 (which is newer than nissli-15.0.0-f1.x86_64) is already installed
	package labview-2015-rte-15.0.1-4.x86_64 (which is newer than labview-2015-rte-15.0.0-4.x86_64) is already installed
	package labview-2015-rte-15.0.1-4.x86_64 (which is newer than labview-2015-rte-15.0.0-4.i386) is already installed

Installation completed.

Do I need to downgrade to LV2015?

 

0 Kudos
Message 1 of 9
(2,406 Views)

This NI Whitepaper deals with a similar issues, but only in establishing there is a problem (absence of dependency) in my case.

0 Kudos
Message 2 of 9
(2,405 Views)

Update/plot thickens.

I uninstalled ALL NI software from SL, then installed 2015 RTE, then NI-VISA. I still have the same error when running Visa Configure:

# visaconf
Can't load library /usr/local/lib/liblvrt.so.15.0
libXinerama.so.1: cannot open shared object file: No such file or directory
To download the LabVIEW Runtime engine, go to http://www.ni.com/rteFinder?dest=lvrte&version=15.0&platform=Linux&lang=en
0 Kudos
Message 3 of 9
(2,373 Views)
Solution
Accepted by sonobat

Hopefully documenting here will help someone out in the future. If you have the same bug post install here is what you do:

 

 

# yum install libXinerama.so.1
# yum install libGL.so.1
# reboot

 

(from the VISA FAQ)

"Some action is required to make the existing NI-VISA installation
work for the new kernel. After upgrading your kernel, run
updateNIDrivers utility as root. The utility is located in
/usr/local/bin."

 

and bam. works.

Message 4 of 9
(2,366 Views)
Highlighted

@sonobat wrote:

After successfully installing Labview 2016 and NI-VISA on a new SL machine, I attempt to run Visa Configuration and get the following:

 

$ visaconf
Can't load library /usr/local/lib/liblvrt.so.15.0
libXinerama.so.1: cannot open shared object file: No such file or directory
To download the LabVIEW Runtime engine, go to http://www.ni.com/rteFinder?dest=lvrte&version=15.0&platform=Linux&lang=en

The package depencies are broken.

Is this software even properly tested before release ?

Those kinds of errors can be easily catched fully automatically in a proper CI setup (which is pretty trivial). Does NI need lectures in the essential software engineering basics ?

 

Linux Embedded / Kernel Hacker / BSP / Driver development / Systems engineering
0 Kudos
Message 5 of 9
(2,091 Views)
Solution
Accepted by sonobat

Thanks a bundle @sonobat. I finally go this software abomination running on the ancient 2.6 kernel with 32 bit libraries bloating my system. If one claims "we supports Linux distro X of version Y and we pr0vides mighty INSTALL script" that the resulting install would just, you know, be able to execute... Obviously NI hired an intern to do their dirty Linux support and no-one bothered with testing his crap before shipping.

 

So for anyone in the future tasked with getting NI VISA 17 running in some kind of Linux distro: here's the missing NI dox.

 

Install eg. Scientific Linux 6.9 64 bit (I used VirtualBox, have not tried yet on real hardware)

Add yourself ('myself' below) to /etc/sudoers for sanity:

% su -
% echo "myself ALL=(ALL) ALL" >> /etc/sudoers
% exit

Install a tool chain:

% sudo yum install -y gcc

<VirtualBox detour>

For VirtualBox: install kernel-devel before installing the guest additions

% sudo yum install kernel-devel kernel-devel-2.6.32-696.1.1.el6.x86_64

Install the VirtualBox guest additions, google if unsure.

</VirtualBox detour>

 

Mount the ISO (who delivers software on ISOs nowadays anyway?)

% sudo mount -o loop NI-VISA-17.0.0.iso /mnt

Install missing dependencies, including a 32 bit glibc and friends 😕

% sudo yum install -y glibc.i686 libstdc++.i686 avahi libXinerama.so.1 libGL.so.1

Finally, run NI's brilliant INSTALL script:

% sudo /mnt/INSTALL

Rebuild kernel modules:

% sudo /usr/local/bin/updateNIDrivers

Reboot. Once back:

% visaconf

You should get not complaints but a rather 90s looking GUI. Now I only have to wait for my instruments to arrive and see if this thing actually works.

 

Message 6 of 9
(1,672 Views)

Well done! moving solution tag to you for greater details. Please post reply after instrument connection misery, I'm quite curious how that will pan out.

0 Kudos
Message 7 of 9
(1,477 Views)

Thanks a bundle @sonobat. I finally go this software abomination running on the ancient 2.6 kernel with 32 bit libraries bloating my system. 

OMGs. 2.6 - seriously ?

 

If one claims "we supports Linux distro X of version Y and we pr0vides mighty INSTALL script" that the resulting install would just, you know, be able to execute... Obviously NI hired an intern to do their dirty Linux support and no-one bothered with testing his crap before shipping.

ACK. I've heared rumors, there's only the one guy ... 

 

So for anyone in the future tasked with getting NI VISA 17 running in some kind of Linux distro: here's the missing NI dox. 

Well, just did some little experiments ... turned out this silly installer has even more bugs:

 

  1. checks for rpm, finds none (in my case correct - rarely have rpm on debian), decides to use its own one, but is to stupid to call it - the variable holding the command isn't set.

    => why that strange game at all ? why not just fail w/ proper error message if rpm isn't installed ?
  2. wrong install location - /usr/local/ prefix is meant for things that are NOT installed via package manager !
  3. applications (especially shell scripts) should NEVER open legacy devices  /dev/stdio, /dev/stderr, /dev/stdin, unless you know exactly what you're doing and *want* it's specific - non-portable - semantics ! Most likely it's just WRONG.

    guess what this is doing:
    for i in 0 1 2 3 ; do (echo foo > /dev/stdout) >> OUT ; done ; wc -l < OUT
  4. the kernel code is so extremly broken and dangerous ... dont even wanna talk about this. elementary school level. whoever wrote that crap has absolutely no idea about linux kernel development.

The quality is so horrible that any serious CTO would strictly prohibit running this code in his organisation and doesn't wanna be held personally responsible for gross fault.

Linux Embedded / Kernel Hacker / BSP / Driver development / Systems engineering
0 Kudos
Message 8 of 9
(1,438 Views)

Thanks for the info. Been beating my head against the wall trying to get this thing working.

0 Kudos
Message 9 of 9
(729 Views)