"How it works" Chapter 4 README.md

[Student007]

How it works

Vagrunt up

The vagrant up command installs the download manager axel and the version control client git on top of a prepared vagrant box (ubuntu/trusty64). Further it downloads and extracts the mipsel-toolchain (276 MB) (Toolchain for cross-compiling which uses uClibc for RISC processors)

Bash Script compile.sh

This script first logs into the VM and causes installing NVM and NodeJS 4.2.1 if missing. Further node-gyp installs node development header files for the corresponding NodeJS version.
Now pre-gypify compiles pre-gyp into non-gyp modules. The package to compile now is installed without pre-install and pre-publish scripts (npm install --ignore-scripts). Afterwards node-pre-gyp build package creates the compiled binary module for the Tessel 2 which is using a RISC Processor Mediatek MT7620n.
Finally it syncs the compiler output via ssh to your current ./out

Binary Server

• The /lib/server.jsis starting the /lib/package-listener.js and a /lib/web-server (http://0.0.0.0:3000) which shows package names from the local ../knexfile database.
• The /lib/package-listener.js is is watching (listenForChanges()) https://skimdb.npmjs.com/registry/ for changes by calling check-for-gyp. It uses concurrent-couch-follower to ensure every packages.json with a binding.gyp is completely checked (in parallel for up to 20 docs at the moment)
• /lib/check-for-gyp.js returns a promise which checks whether a binding.gyp exists in the doc (package.json)
• /lib/package-url.js is fetching and parsing the packages.json from npmjs.com
• db.js sets up the local kexfile.js

[Student007]

@reconbot is the /lib/package-builder.js replaced by content/calling of compile.sh or a rework ?
My current understanding is, you call t2-identify.js packageName and than things happen like described above.