Why? What? How?

[nelsonic]

This repo was born out of the need identified in: https://github.com/dwyl/github-reference/issues/15

Why?

As a person who uses GitHub as their "single-source-of-truth" I need a backup of GitHub issues so that I can work "offline". (either when I have no network or GH is "down")

What?

Backup GitHub issues.

• a script that lets you login to GitHub using OAuth
• Use GitHub REST API to retrieve all issues
• Store issues as "flat files" in subdirectories corresponding to the GitHub URL e.g: if the issue is dwyl/github-reference/issues/15 store it as dwyl/github-reference/issues/15.json
• It does not need to "scale" it just needs to work. read: http://paulgraham.com/ds.html

Note: initially we do not need a Web-UI to let people backup their issues, but later we could build UI to make it more useable.

How?

I have a strong preference for using Elixir for this because we are transitioning all our projects to Elixir, but given that this script does not require parallelism/concurrency it's "OK" to write it in which ever language you are most comfortable using.

[SimonLab]

Some questions and comments to try to get a better idea on how to implement the backup:

• Do we need to implement OAuth, maybe a first version can get directly the issues from the API without authentication. I think the only reason why we would want to use OAuth is to have a better number for the requests limit, we might need to know how many requests the script needs.

• Will this script run locally or on a server? In both case we need to save the issue json files asked locally.

• If we want to work offline we migth need to think how to synchronise the offline backup of issues with Github but this might be a feature needed for the next version not now.

Let's try with Elixir! The "Programming Elixir" book has already a nice example which explains how to display the first n most recent issues of a repo.

[nelsonic]

@SimonLab if we use the API without OAuth there will be a limit for the number of requests per minute/hour. you're correct that this will work, it will just mean the person has to wait for the script to run over a number of hours instead of working in a few minutes.

Also, @jruts managed to get a GitHub Web Hook with Elixir set up quite quickly: https://github.com/jruts/git_webooks_elixir so might be worth doing a bit of googling.

[jruts]

@nelsonic that is just to play around, it is far from complete :) Elixir is too much fun not to play around with it. I do think it is rather easy to implement this mini project. If you need a ready to use github lib I would recommend tentacat.

[nelsonic]

@jruts thanks for the tip! https://github.com/edgurgel/tentacat looks really good! 😍 (hope your day is going well...)

[jay-meister]

@nelsonic is this mini project no longer desired now you have seen tentacat?

[nelsonic]

@JMurphyWeb I think we just need to write a script that uses Tentacat to fetch all the issues for an org's repos.

[nelsonic]

If we could solve this issue: https://github.com/dear-github/dear-github/issues/129 the world (of GitHub) would be a better place!

[Cleop]

Closing as 'Done' in the Projects board.