hexo init blog

I’ve been wanting to start a blog for a long time now. Today I’m pulling the trigger on that with a simple hexo init blog. Well, it wasn’t that simple, so I feel like it’s worth talking about a few of the complications I had.

Hexo, my chosen blogging framework, has made some interesting decisions. Running tree -aACL 1 shows the directory structure for my new Hexo blog.

.
├── _config.yml
├── db.json
├── .deploy_git
├── .gitignore
├── node_modules
├── package.json
├── public
├── scaffolds
├── source
└── themes

6 directories, 4 files

Hexo keeps its generated static files in the public directory. When deploying with git, as I choose to do, it uses the hidden .deploy_git directory to keep changes to public under version control. When generating static files Hexo obliterates anything in the public directory it doesn’t know about. This means that even if the user wants to add some static files of their own they’ll just be baleeted when Hexo clears out the .deploy_git directory and copies the public directory in.

This isn’t a big deal if you don’t care about having any static files not generated by Hexo on your site. It just so happens that I’m very keen on having my proof for Keybase. In order to prove ownership of a website Keybase requires that you keep a file called keybase.txt or .well-known/keybase.txt from the root of your site.

So, how do you serve a custom static file when your blogging framework is intent on keeping you from doing so? I suppose I could do the responsible thing and submit a pull request upstream for a configuration option to preserve certain files, but I’m much too lazy for that. Instead I’ve enlisted nginx for a band-aid solution.

location =/keybase.txt {
root /srv;
}

Most requests to my site pass through nginx to my default root, but any requests for keybase.txt get served out of a separate directory where it safely sits.