Set up CI to auto-run progress.sh

[Roger Oba]

This could probably be GitHub Actions, since it's free and easy to use :)

[Stefan Herold]

I found this and this article. We could use the same principle to generate the README.md instead of editing it all the time. Steps would be the following:

• instead of contributing to the readme we create a new folder 2020
• pre-populate it with md files named session_number.md
• Fill the files with the same stubs you created earlier this week
• once a PR is merged README.md is re-generated using a GitHub action

Advantages:

• No more manual TOC editing
• No more conflicts
• Clean file structure
• Older years could be part of the default branch and don't have to be archived on another branch
• Progress could be added automatically

What do you think? I also have use for these separate MD files in another side project...

[Roger Oba]

Interesting approach, I had thought of something similar but thought that it would be an overkill.

I like the pros you raised, though. I think the only con is implementing it (time consuming implementation).
So, comparing "leaving things as is" to the advantages you raised:

• ToC editing would still be manual, but not a big deal imo.
• Conflicts are already minimum AFAIK, since the stubbing of the sessions
• I like the clean file structure you proposed, better than what we have now.
• We actually don't need the 2019 branch, we can create a 2019_sessions.md with the 2019 branch's README and make it live in master, and README.md would always contain the latest WWDC (2020 for now, but in june 2021 it'd be 2021's).
• Progress could be added via CI without all that hassle, see #59 :) we'd just have to plug this in CI

[Stefan Herold]

I think I'll face the effort. It will be worth it in the end 👍 I'll make it my weekend project ^^

[Stefan Herold]

@rogerluan which algorithm did you use to create the site link for the toc entries (#session_header)

[Roger Oba]

I just wrote this:

func normalizeTitleToHyperlink(_ title: String) -> String {
    return title
        .lowercased()
        .replacingOccurrences(of: "★", with: "")
        .replacingOccurrences(of: " ", with: "-")
        .replacingOccurrences(of: "'", with: "")
        .replacingOccurrences(of: "’", with: "")
        .replacingOccurrences(of: "\"", with: "")
        .replacingOccurrences(of: ":", with: "")
        .replacingOccurrences(of: ",", with: "")
        .replacingOccurrences(of: "?", with: "")
        .replacingOccurrences(of: "&", with: "")
}

It's not extensive, but worked out for all the nearly 200 video session entries. To validate the entries, I scanned all the output:

let normalizedLine = normalizeTitleToHyperlink(line)
var set = CharacterSet.alphanumerics
set.insert(charactersIn: "-")
if normalizedLine.rangeOfCharacter(from: set.inverted) != nil {
    print("## \(line)\t\t\tBroke and printed: \(normalizedLine)")
    didFailForAtLeastOneEntry = true
}

[Roger Oba]

Do you have Telegram? reach me on https://t.me/rogerluan @Blackjacx :)

[Stefan Herold]

I will tomorrow, late now here :-)

If you have time you can have a look over PR #62
It resolves 2 of the issues :-) Would be cool to have it cross checked before I merge it ^^
After this I'll continue watching sessions like crazy ✌️

It is better to read the commits instead of all the changes at once

[Stefan Herold]

Okay the PR is merged. I'll close this when it is confirmed that the readme generation is working

[Stefan Herold]

Well it rocks ✌️