The mechanics of what's here...

Published: Fri 27 November 2020
Updated: Wed 23 December 2020
By Michael Scharf

In Meta.

Q: "Which static-site generator makes you happy as a Python developer / what did you end up using?"

I looked at js-based solutions like ghost (too much), and the great Go-based Hugo (maybe after I finally learn Go). In Ruby-world, there is the cool-looking but seemingly unmaintained toto (last commit in 2013), and the major player: Jekyll.

Jekyll seems like the obvious choice: it is the force behind GitHub Pages, which I have actually used for prior projects. And the amazing-looking staticman, which actually seems to solve the problem of comments for static sites, is built for it.

The problem, for me, with the Jekyll-GitHub Pages-staticman setup is the dark side of io. Even if you are using your own domain with Github pages, it seemingly relies under the hood on a .io domain, which I don't want to engage.

And so, since I also didn't particularly want to get involved with Disqus, I decided to forget about comments for now and go with Pelican. It's Python, it's battle-tested, and it's simple to host via a cloud bucket. And that "Pelican" turns out to be an anagram of calepin, which is French for "notebook," made it that much nicer.

Which cloud bucket? I've ended up on Google Cloud Platform for my personal projects. The reason has been, up to this point, Firebase (which was first recommended to me by my Sensei). When you're writing a few pages by hand, and/or need a db, there doesn't seem to be an easier way to host a site. That's not really what this is, though, so when I thought about this site, I thought first: Google Domains and then: Google Cloud Storage must, by now, have the same functionality as GitHub Pages. That second thought looked to be correct, and once I found this tutorial for static site hosting on GCP, I decided to try it.

For my first Pelican-based post, I read the quickstart, and followed the directions. My markdown is not particularly fast, but there are cheat sheets. (I won't post one because I haven't settled on any.)

I bought the domain and slowly followed the hosting tutorial directions. Once I had completed all the steps, I tested via

$ pelican content

followed by

$ pelican --listen

and everything looked OK.

I uploadled the contents of the entire Pelican-generated output folder and all its subfolders to the cloud bucket via the bucket's UI. (That is, not output itself as parent: yourbucketname takes the place of output-as-parent-folder.) And then I tried it, and my hello was live automagically.

To publish this post, will take an incremental step toward CI by using Google Cloud's version of rsync. From this post, I learned the following: "The -m flag accelerates upload by processing multiple files in parallel, and the -c flag avoids re-uploading unchanged files." The -r flag recurses through the subdirs. The -d flag, the doc says, "provides a means of making the contents of a destination bucket or directory match those of a source bucket or directory."

So the final command, having installed the gsutil tool is:

$ gsutil -m rsync -r -d /Users/Mike/projects/pythinkloop/output gs://pythinkloop.com

and if it works, this should show up...

(it worked after I did this within the env)

Influences and resources

social