Review questionnaire pipeline

[edipetres]

With regard to my meeting with Bue

• questionnaire-engine
• questionnaire-models

• 143 settle on design decisions

[buep]

Feedback

engine pipeline

• The release step I prefer always is manual, I don't want a new questionnaire-engine release just because I edit a typo in the readme. So make the release job that pushes manually executed so someone with access can click run to create the release.
• version is hard-coded in the build step, this means you will edit the build step when you want a version bump. Confusing for other why the build definition file is changed, when build definition is not changed. The version number needs to come from a file, e.g. version.sh with major minor patch number so when one needs to change version that file is the one to edit. Appeding the CI build number is good as you already do
• the tag you use on docker hub for release number, should also be present at the same git sha in github, so remember to tag the github repo also
• the docker hub repo misses readme description, it should show the same content at the github readme and be automatically updated or posted there - all the information should be alligned
• I can't find the credentials in our password management system, when I search for AWS there is nothing related to questionnaire.... so either point to specific token or a better text to search for, or explain more precise in the readme which account is used. e.g. for example you only write "AWS_ACCESS_KEY_ID | Acces key for the AWS account the pipeline deploys to" but that doesn't tell me what to look for in our internal list
• I'm missing test steps, that shows the container can be started and serve a questionnaire with the demo content - else we risk publishing a container engine that doesn't work

content pipeline

• same observations as above with version number
• same things with docker hub readme and description, should be automated and contain proper content
• also need to explain why there is an AWS version number also
• missing a test step that make sure the container can run and serve the questionnaire locally before deployed to AWS
• all steps here in the content pipeline is fine to be fully automated and deployment to happen automatically

[buep]

On a general note... think we should create a questionnaire-meta repo with much of the generic content that explains the hole overview about engine and model... and then point to the concrete ones.

Let's discuss some improve docs - there are so much fine documentation you create, we just need to structure it a bit further for others to understand it.

[buep]

one last thing.... do we not have a Praqma circle ci organisation? this looks strange (see picture)

[edipetres]

@buep not sure what's wrong on the picture, I use the pipeline with the Praqma organization account on CircleCI

[buep]

Shouldn't the circle ci organisation be called Praqma? Or maybe it is because I'm not member?

[edipetres]

It is called Praqma for me - so maybe you need to be a member.. but that Bitbucket name is weird.

Closing this as we created new issues for all the points here.