The working, but not great, Jekyll comment plugin development story
- 23 Feb 2019: Post was created (diff)
I was building a comment API to accept comments to my Jekyll blog (this blog). I wanted to be able to accept comments and get them back out again. This had me planning for two API endpoints:
When I was done and things worked out alright, it was time to put into production on my web server. First issue I stumbled into, (which is not the first time, at all) was to have the Node.js version available on my server. My server, however did not have any modern versions of Node.js available in the standard repositories. But I didn’t want to pollute the server with new repos as I did not want to burden my application with too much deployment steps.
I ended up packing it into a Dockerfile, as I already had Docker running on the server. Then I had to expose a local directory to have a place to store the sqlite database file. I didn’t really like this too much either. Something felt wrong.
When everything was up and running, and the Jekyll generator plugin worked as it should too, I realized this wasn’t what I wanted at all. I was indeed very happy with finishing this small project to the end, but I was somewhat surprised at how wrong everything about this solution felt.
I had a static site whos strength was that it was generated and thereby only consisting of plain files stuffed into a git repo. This was beautiful. But now, I had introduced an external dependency, of which could not live in the same repo. This was indeed not what I wanted.
Now I’m starting over again with a Python 3.5 HTTP back-end which spits out comment files as they come in, then adds them to the repository. 3.5 because it seems to be included in most of the mainstream distros these days. I imagine things will turn out better this time around. Let’s see…
- The code to the Node.js SQLite solution
- Jekyll comments generator plugin that uses
curlto download comments
The code for the Python 3.5 plain file solution may just pop up some day