RFC: separate the website to repos

[Roger-luo]

I tried to revise the implementation a little bit, and I'm wondering if it'd be easier to maintain if we separate the website to repos under this org. The new org would have the following repos:

• physicsml.github.io: a shiny front page with links to the following
• awesome-physicsml: a pure markdown list with a gh-pages that generates from the markdown, and it follows the github trending awesome list tag
• blog: contains a static site generator like jekyll/hexo for people posting blogs.

I found most static site generators are made for blogs, which doesn't actually have nice theme for a frontpage, and several single pages for lists of papers in each year, which make it hard to choose between a nice theme and customization.

The first two might just be some pure html + custom rendering scripts, which make it much simpler, and for the blog, we could have it when there are some blog posts. It would be even better if someone could donate a domain, like physicsml.org etc. so we could just have blog.physicsml.org, awesome.physicsml.org, etc. (maybe even events.physicsml.org!)

PS. I don't have access to the org, so someone might want to let me in first...

cc: @AnnaGolubeva @DannyKong12 @mbeach42

[everthemore]

The domain itself is available and costs only approx $15/year. I could offer to host it on my Hostgator account (which is hosting a few others atm too)? But perhaps having it hosted through a more independent place (i.e. university or institute?) makes more sense.

The nice thing about this format is that, in principle, people could suggest papers through pull requests. The downside is that a large fraction of those people might be unfamiliar with pull requests, so a submission form would be useful for them?

[Roger-luo]

I can make a github bot for the paper submission, my proposed workflow would be:

• open an issue with certain format (we can make an issue template for this)
• some core member can review the issue, while everyone could comment, then type approve which will trigger the bot later
• the adim will run the robot script once a week/a period which will collect the passed submission
• markdown file gets updated

A local script would be easy to make with PyGithub or Julia Github.jl, a lot CI bots are using that. It goes more like an open source journal lol. If someone want to donate a sever we can then set up a real Github bot and make all things automatic.

The point of splitting the repo is mainly because:

1. static index is not updated frequently
2. it would be better to make the paper list repo as simple as possible, so it can be also updated by hand easily. (in the proposal, it only contains a few markdown files, and a script generates the static html, and maybe a css file)

PS. I believe using github is something people need to adopt today. It is not hard to learn, and we can still update this manual like before (by receiving emails).