Set up azure-pipelines to automatically build and publish the documentation

[datapythonista]

After #31 is complete, we should set up a continuous integration system (azure-pipelines) to automatically build the documentation to html, and publish it in a website, every time there are changes.

[bhavaniravi]

I wanted to learn CI for a long time. I would love to work on this.

[sparalic]

I would love to work on this with @bhavaniravi as well.

[datapythonista]

Great, let me know if you have any question.

[bhavaniravi]

Hi, @datapythonista I understood how azure-pipelines work with GitHub and tried to set one up with my personal account pointing to the forked repo.

I have a couple of questions

1. Are we going to use a common MS account to set these things up
2. I am looking at azure pipelines in pandas repo and removed the unwanted parts not sure if I should remove Checks job as well
3. The last stage of the pipeline is to push the built doc to a repo, is it a separate repository or the one we are working in right now.

For others to learn what I have done

1. https://www.azuredevopslabs.com/labs/azuredevops/github-integration/
   2 Followed InstallSSHKey@0 from the pipeline file

[datapythonista]

You can decide about how to manage the permissions / account. You can create a new Azure account and share the credentials with the group, or create with your user, and give admin permissions to some or all people. Not sure if there is any other option.

I'd start very very simple. I'm sure you can start with already complex jobs copied from pandas, and have useful things, but the thing I care the most at this point is that everybody is able to understand what's the CI, what it does for us, how is set up... And if the PR is more than few lines, it'll be difficult for the rest to follow.

Does this make sense?

[bhavaniravi]

Understood. Let me reiterate and get back.

[bhavaniravi]

Now that #83 is merged, back to working on this

[bhavaniravi]

I am kind of done with setting up a basic pipeline that runs the make file successfully.

Here is where I am stuck, where do we store the generated HTML files.

1. Are we going to have a separate docs repo?
2. Are we going to push to the same repo?
3. Is there a way to update code directly from pipeline instead of pushing?

[datapythonista]

You can see what we do in pandas here: https://github.com/pandas-dev/pandas/blob/master/azure-pipelines.yml#L120

Basically we have a specific repo that we publish with github pages, and that we overwrite every time. There are surely other options, including pushing the same repo. Ideally we'd like to have 3, I contacted azure pipelines few times to see if they can publish an artifact as html, but haven't heard from them, so I don't think that's an option.

[bhavaniravi]

guess what? I could successfully push the HTML docs to a separate repo.
Now, can we create a pandas-mentoring-docs repo under python-sprints?

[datapythonista]

Sure, I think I already made you an admin of the organization, you should be able to create it yourself. :)