[nan-668] add verification bash script to publish packages

[khaliqgant]

Describe your changes

The goal with this PR is to mimic a publish process of the cli with the code that is in review to ensure a working cli that is in a remote source. This is accomplished by building and publishing the packages to the Github registry and finally using the cli to verify that it works expected. This will protect against instances where a faulty cli is published. This will be especially useful to verify that this PR: #1918 works as expected with packaging the utils package. Additionally, as we refactor the cli to move code out of shared and reduce the package size of the cli this will ensure things still work as expected.

Issue ticket number and link

NAN-668

Checklist before requesting a review (skip if just adding/editing APIs & templates)

• I added tests, otherwise the reason is:
• I added observability, otherwise the reason is:
• I added analytics, otherwise the reason is:

[linear]

NAN-668 Add cli test to ensure bad versions are published that won't compile properly

[TBonnin]

isn't the cli published by running the publish.sh script in the previous step? why do we need to publish it again?

[khaliqgant]

It attempts to publish it but fails because it is not scoped under our organization @nangohq which is why I rename the package and publish it separately.

nango/.github/workflows/cli-verification.yaml

Lines 39 to 45 in f9448fc

	- name: Publish the cli privately under the correct scope
	working-directory: packages/cli
	env:
	NODE_AUTH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
	run: |
	jq '.name = "@nangohq/cli"' package.json > temp.json && mv temp.json package.json
	npm publish --access public

[bodinsamuel]

why not doing that directly then?
Also doesn't it will publish @nangohq/node and @nangohq/shared every time for nothing?

[khaliqgant]

why not doing that directly then?

Easier to reuse the script which bumps the references and packages correctly.

Also doesn't it will publish @nangohq/node and @nangohq/shared every time for nothing?

Shared needs to be published, node is published again and not used as is frontend but there is no harm in publishing those packages.

[bodinsamuel]

why not doing that directly then?
Also doesn't it will publish @nangohq/node and @nangohq/shared every time for nothing?

[TBonnin]

It is a bit of an hacky solution but since it is contained in a single github actions I am fine with it. We can improve it in the future if the need arises

[khaliqgant]

It is a bit of hacky solution but since it is contained in a single github actions I am fine with it. We can improve it in the future if the need arises

Agree it is a bit (read: very) hacky but also agree if it causes problems we can improve or rip it out even. The hope is that it saves us in the future. Time will tell if that is actually the case.