How to access files

I am currently using the Jetson Nano 2gb through the headless setup but am curious on if there is a way I could open up and manually view the files within there like image files.

Do you mean you would like to login Jetson Nano and edit the files directly in console window?

Or do you mean view graphical images like a JPEG?

If so, and you are using another Linux computer with a GUI, then you can tell ssh logins to forward GUI requests to your host PC. The Jetson’s GPU would not be used as the GUI requirements would forward to the host PC. Let’s say you were going to just open gimp on the Jetson by logging in to account “nvidia” (adjust for your actual login name) on the Jetson (I’ll pretend it is IP address, but adjust to the actual IP address):

ssh -Y nvidia@
gimp &

In the above the “-Y” options (also “-X” works, it differs in security) says X server events forward to your host PC via the ssh (the Jetson GPU is ignored). If your host PC has all of the required libraries and display components to function with gimp in that example, then it will display on the host PC. Similar for any GUI program.

One can invoke a single command and keep the ssh running for just that command (versus interactive shell):
ssh -Y nvidia@ gimp
(this will run on the Jetson but display on the host PC until gimp is exited)

If you need the GPU of the Jetson to be used, e.g., to monitor what the Jetson is doing, then you need a virtual X server. ssh forwards the events which trigger X server graphics, whereas a virtual desktop keeps the events on the Jetson, but forwards the results of those events to a remove virtual desktop client running on another computer. Actually, it is possible to run the client and server on the same computer, but not many people would do that.

One advantage of the virtual server is that all requirements are on the Jetson itself, and the GPU is not bypassed. Also, for X forwarding, the host PC must have all of the X server and library components running at the proper release version; a virtual desktop program has no such requirement. Windows for example does not run an X server, and if you have a desktop client, then any computer with that client can run the Jetson GUI from a remote location. The entire desktop appears.

A virtual desktop server though is far more complicated to install and set up. You’d be advised to back up your work before trying. It involves messing with the NVIDIA GUI driver, and if you unintentionally set up a virtual server with a Nouveau driver, then CUDA and many other functions will cease to work.

More like be able to open and look at any of the code within files of the Jetson Nano, with like numbers on the left side correlating to each line of code, in the way most coding websites UI look like. I am attempting to create code through python but am not sure how to create such a file.

FYI, there are a lot of choices, and much of it depends on your tastes. A Nano might not have much hard drive storage (e.g., depends on SD card size), so that too changes things. I tend to just use vim, but this has a steep learning curve. Often I use the KDE application kate, but on a non-KDE system this probably means adding a lot of support files. Most of the time one would use ordinary system package tools to install, but there are some packages you have to get separately (namely the Microsoft VSCode), and others use the Python “snap” program to install (then you get to the complication of whether to use the Python 2 or 3 version).

An example to update your system before starting:

sudo apt update
sudo apt-get upgrade

(not necessarily required, but often good to do once before developing)

One editor I know of is spyder3. You could search like this:
apt search spyder3

This would be an absolutely enormous list, and so I’ll add a “pager” so it doesn’t scroll away:
apt search editor | less
(then you can scroll up and down in the list; piping to “less” is the pager)

To install spyder3:
sudo apt-get install spyder3

Here’s a URL on VS Code (use the Ubuntu instructions):
(if you run into multiple versions, e.g., not via package manager consulting a repository, then you’d want the arm64/aarch64 version).

Also, some editors use more RAM than others.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.