Ulysses’ trip | Guide

How to use JavaScript in GitHub Actions without committing `node_modules`

2024-11-15T16:34:55-08:00

I was creating my first GitHub Action (it is now working, which you can check out here). I am familiar with JavaScript, so I chose to create a JavaScript action. According to the official documentation, I need to do one of the following to make my JavaScript code correctly import 3rd-party modules:

commit the node_modules to the repo, or
use @vercel/ncc to bundle everything into a single script.

I think both of these methods are not good practices. Especially, for my case, the way with @vercel/ncc is not feasible because some package uses require.

The best way to handle this is to make GitHub cache and install all the dependencies for the user every time the action is run. This can be made possible by creating a composite action. Here is a snippet in action.yml:

          
            runs:

              using: composite

              steps:

                - name: Setup Node.js

                  uses: actions/setup-node@v4

                  with:

                    node-version: '20'

            

                - name: Get cache key and path

                  id: lock

                  run: |

                    cd $GITHUB_ACTION_PATH

                    node <<'JAVASCRIPT'

                      const fs = require('fs');

                      const os = require('os');

                      const sha256 = require('crypto').createHash('sha256');

                      const hash = sha256.update(fs.readFileSync('package-lock.json')).digest('hex');

                      fs.appendFileSync(process.env.GITHUB_OUTPUT, `key=${os.platform()}-${os.arch()}-grs-modules-${hash}\n`);

                    JAVASCRIPT

                    modules=$(pwd)/node_modules

                    command -v cygpath &> /dev/null && modules=$(cygpath -w $modules)

                    echo "modules=$modules" >> $GITHUB_OUTPUT

                    cat $GITHUB_OUTPUT

                    cd $GITHUB_WORKSPACE

                  shell: bash

            

                - name: Restore cache

                  uses: actions/cache/restore@v4

                  id: cache-restore

                  with:

                    path: ${{ steps.lock.outputs.modules }}

                    key: ${{ steps.lock.outputs.key }}

            

                - name: CI

                  if: steps.cache-restore.outputs.cache-hit != 'true'

                  run: |

                    cd $GITHUB_ACTION_PATH

                    npm ci

                    cd $GITHUB_WORKSPACE

                  shell: bash

            

                - name: Save cache

                  if: steps.cache-restore.outputs.cache-hit != 'true'

                  uses: actions/cache/save@v4

                  with:

                    path: ${{ steps.lock.outputs.modules }}

                    key: ${{ steps.lock.outputs.key }}

            

                - name: Run

                  run: node $GITHUB_ACTION_PATH/index.js

                  shell: bash

I manually use actions/cache to cache the node_modules directory although actions/setup-node has caching functionality. This is because it is not good yet. Also, because github.action_path ends with /./, it can cause some problem due to the relative pathing. My workflow here works around those problems.

The hash in the cache key is calculated using JavaScript because there is no sha256sum command on macOS and Ubuntu runners.

Letting people know when you are asleep

2024-02-21T02:26:45-08:00

While I have been trying to respond whenever people reach out to me, it is impossible to be available 24/7 due to the simple fact that a human has to sleep to be alive. While this is annoying, it is also a fact that I cannot change.

Therefore, a natural idea that comes to my mind is to simply let people know when I am asleep so that they do not expect me to respond immediately.

Discord and GitHub have been among my most used platforms for some time, and they both have a feature to let users set a custom status (with a custom text and an emoji). This opens up a possibility of using a program to automatically set the status to indicate that I am asleep.

For Discord, this is as simple as invoking a REST API (Notice that this is against Discord’s ToS):

          
            # Set sleeping status

            curl -X PATCH \

            	-H "Content-Type: application/json" \

            	-H "Authorization: YoUr.DiScOrD.ToKeN" \

            	-d '{"settings":"WhwKBQoDZG5kEhMKC1NsZWVwaW5nLi4uGgTwn5i0"}' \

            	https://discord.com/api/v9/users/@me/settings-proto/1

            

            

            # Clear sleeping status

            curl -X PATCH \

            	-H "Content-Type: application/json" \

            	-H "Authorization: YoUr.DiScOrD.ToKeN" \

            	-d '{"settings":"WgoKCAoGb25saW5l"}' \

            	https://discord.com/api/v9/users/@me/settings-proto/1

If you wonder what those base64-encoded strings mean, they are actually the binary contents of protobuf-encoded messages. See this .proto file for the definition of the message structure.

API v8

Using nmcli to connect to eduroam in UCSB

2024-01-08T01:03:36-08:00

Save this certificate to some file, say /YOUR/PATH/TO/ca.pem:

Then, run

          
            nmcli con mod eduroam 802-1x.eap peap

            nmcli con mod eduroam 802-11-wireless-security.key-mgmt wpa-eap

            nmcli con mod eduroam 802-11-wireless-security.proto rsn

            nmcli con mod eduroam 802-11-wireless-security.pairwise ccmp

            nmcli con mod eduroam 802-11-wireless-security.group ccmp,tkip

            nmcli con mod eduroam 802-1x.ca-cert /YOUR/PATH/TO/ca.pem

            nmcli con mod eduroam 802-1x.phase2-autheap mschapv2

            nmcli con mod eduroam 802-1x.anonymous-identity anonymous@ucsb.edu

            nmcli con mod eduroam 802-1x.identity YOUR_EDU_EMAIL_ADDRESS

            nmcli con mod eduroam 802-1x.password YOUR_PASSWORD

Finally, you can connect to eduroam now!

This may also apply to eduroam in other campuses, but I haven’t tested it yet.

How to self-host Overleaf (ShareLaTeX)

2023-09-29T11:11:38-07:00

Install Docker and NGINX before you go. Replace every overleaf.your-domain.com with your own domain name.

Set up NGINX

          
            server {

            	listen 80;

            	listen [::]:80;

            	server_name overleaf.your-domain.com;

            	return 301 https://$host$request_uri;

            }

            server {

            	listen 443 ssl;

            	listen [::]:443 ssl;

            	server_name overleaf.your-domain.com;

            	ssl_certificate /path/to/your/cert/fullchain.pem;

            	ssl_certificate_key /path/to/your/cert/privkey.pem;

            	location / {

            		proxy_set_header Host $host;

            		proxy_pass http://localhost:8444; # use a port you like

            		proxy_redirect off;

            

            		proxy_http_version 1.1;

            		proxy_set_header Upgrade $http_upgrade;

            		proxy_set_header Connection $connection_upgrade;

            		proxy_set_header X-Scheme $scheme;

            

            		proxy_buffering off;

            	}

            	location ~ /.well-known {

            		allow all;

            	}

            }

Set up ShareLaTeX

Refer to: quick start guide.

Run the following:

          
            OVERLEAF_HOME=./overleaf # set to whatever you want; sudo in following if not having write access

            git clone https://github.com/overleaf/toolkit.git $OVERLEAF_HOME

            cd $OVERLEAF_HOME

            bin/init

Notice: You may look at git log -n 1 --pretty=format:"%H". Mine was cc4d01bb46d4e0d7c08124372ff69a4578e7333d. I can guarantee that the following steps work with this version of Overleaf toolkit, but may fail in future versions.

Edit config/variables.env, add the following:

          
            # See https://github.com/overleaf/overleaf/issues/1044#issuecomment-1741289459

            PATH=/usr/local/texlive/2023/bin/x86_64-linux:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin

            

            SHARELATEX_BEHIND_PROXY=true

            # Do NOT set SHARELATEX_SECURE_COOKIE to true, see https://github.com/overleaf/overleaf/issues/388#issuecomment-1741162658

            

            SHARELATEX_SITE_URL=https://overleaf.your-domain.com

            # If you want, set SHARELATEX_APP_NAME, SHARELATEX_NAV_TITLE, SHARELATEX_HEADER_IMAGE_URL, SHARELATEX_ADMIN_EMAIL.

            

            # Email settings, see https://github.com/overleaf/overleaf/issues/816#issuecomment-864665071

            SHARELATEX_EMAIL_FROM_ADDRESS=Overleaf 

            # SHARELATEX_EMAIL_REPLY_TO does not seem to work.

            SHARELATEX_EMAIL_SMTP_HOST=smtp.domain.com

            SHARELATEX_EMAIL_SMTP_PORT=465

            SHARELATEX_EMAIL_SMTP_SECURE=true

            SHARELATEX_EMAIL_SMTP_USER=your.email@domain.com

            SHARELATEX_EMAIL_SMTP_PASS=yourpassword

            SHARELATEX_EMAIL_SMTP_TLS_REJECT_UNAUTH=false

            SHARELATEX_EMAIL_SMTP_IGNORE_TLS=false

            

            # Uncomment this when you want to debug:

            #LOG_LEVEL=debug

            

            # Search for process.env in github.com/overleaf/overleaf to see more options (shame for not documenting them):

            # https://github.com/search?q=repo%3Aoverleaf%2Foverleaf+process.env&type=code

Edit these entries in config/overleaf.rc:

          
            # Match the port in NGINX conf

            SHARELATEX_PORT=8444

Create a file config/docker-compose.override.yml and write:

          
            ---

            version: '2.2'

            services:

              mongo:

                restart: unless-stopped

                container_name: overleaf-mongo

            

              redis:

                restart: unless-stopped

                container_name: overleaf-redis

            

              sharelatex:

                restart: unless-stopped

                #image: sharelatex/sharelatex:with-texlive-full # will be uncommented later

                container_name: overleaf-sharelatex

                stop_grace_period: 10s # see https://github.com/overleaf/overleaf/issues/1156

Run bin/up. Wait for the containers to be up. Then, go to https://overleaf.your-domain.com/launchpad and set up the admin account. Use bin/logs in another shell to check the logs if there are any issues.

Set up TeXLive

Refer to: upgrading TeXLive.

While the containers are up, run bin/shell and run tlmgr install scheme-full. You need to wait for a long time. If you run into the problem of “Local TeX Live is older than remote repository,” you need to run update-tlmgr-latest.sh in the container (after fetching it from CTAN first).

After that, run docker commit overleaf-sharelatex sharelatex/sharelatex:with-texlive-full. Then, edit config/docker-compose.override.yml and uncomment the line image: sharelatex/sharelatex:with-texlive-full. Then, run

          
            bin/stop

            bin/docker-compose rm -f sharelatex

            bin/up

When you upgrade later, you need to re-comment the line in config/docker-compose.override.yml, delete the container, and do the above steps again.