# Docker with GUI output
###### tags: `Docker` `Python`

## environment
HW system: lenovo legion-y530-15ich (i7-8750h, 8G RAM, NVIDIA-GEFORCE-GTX-1050)
OS: Ubuntu Desktop 20.04.2 LTS, kernel 5.8.0-48
Docker version: 20.10.5
Docker image: nvcr.io/nvidia/tensorflow:21.02-tf1-py3

## SOP

### host

$ sudo apt-get install x11-xserver-utils //install x11 application
$ xhost + //let everyone can use x11 application, including docker
//we have to type "xhost +" command after booting everytime

/* added these parameters when you create container: 
-v /tmp/.X11-unix:/tmp/.X11-unix => share host unix graphical output to docker
-e DISPLAY=unix$DISPLAY => change environment about graphic output of docker to let it same as host
-e GDK_SCALE => i haven't studied it
-e GDK_DPI_SCALE => i haven't studied it
*/
$ docker run --gpus=all -it --name x11_tf -v /home/yhwu/docker_volume:/volume -v /tmp/.X11-unix:/tmp/.X11-unix -e DISPLAY=unix$DISPLAY -e GDK_SCALE -e GDK_DPI_SCALE -w /volume nvcr.io/nvidia/tensorflow:21.02-tf1-py3 //for example
/* to split command: 
docker run 
--gpus=all 
-it 
--name x11_tf 
-v /home/yhwu/docker_volume:/volume 
-v /tmp/.X11-unix:/tmp/.X11-unix  => added this
-e DISPLAY=unix$DISPLAY  => added this
-e GDK_SCALE  => added this
-e GDK_DPI_SCALE  => added this
-w /volume 
nvcr.io/nvidia/tensorflow:21.02-tf1-py3
*/

### in container

$ apt update
$ apt-get dist-upgrade
$ apt install build-essential zlib1g-dev libncurses5-dev libgdbm-dev libnss3-dev libssl-dev libreadline-dev libffi-dev python-tk python3-tk tk-dev

## source
1. https://blog.csdn.net/sirobot/article/details/106836332
2. https://askubuntu.com/questions/1224230/how-to-install-tkinter-for-python-3-8
3. https://ngc.nvidia.com/catalog/containers/nvidia:tensorflow