💾 Archived View for perso.pw › blog › articles › webzine-development.gmi captured on 2021-12-04 at 18:04:22. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2021-12-03)
-=-=-=-=-=-=-
Hello readers! I just started a Webzine dedicated to the OpenBSD project and community. I'd like to tell you the process of its creation.
A week ago I joked on an french OpenBSD IRC channel that it would be nice to do a webzine to gather some quotes and links about OpenBSD, I didn't thought it would be real a few days later. OpenBSD has a small community and even if we can get some news from Mastodon, Twitter, watching new commits, writing blog articles about stuff, we had nothing gathering all of that. I can't imagine most OpenBSD users being able or willing to follow everything happening in the project, so I thought a Webzine targeting average OpenBSD users would be fine. The ultimate accomplishment would be that when we release a new Webzine issue, readers would enjoy reading it with a nice cup of their favorite drink, like if it was one's favorite hobby 'zine.
At first I wanted the Webzine to look like a news paper, so I tried to use Scribus (used to make magazines and serious stuff) and make a mockup to see what it would look like. Then I shared it with a small French community and some people suggested I should use LaTeX for the job, I replied it was not great for handling the layout exactly as I wanted but I challenged that person to show me something done with LaTeX that looks better than my Scribus mockup.
One hour later, that person came with a PDF generated from LaTeX with the same content, and it looked very great! I like LaTeX but I couldn't believe it could be used efficiently for this job. I immediately made changes to my Scribus version to improve it, taking the LaTeX PDF version as a model and I released a new version. At that time, I had two PDF generated from two different tools.
A few people suggested me to make a version using mdoc, I joked because it wasn't serious, but because boredom is a powerful driving force I decided to reuse the content of my mockup to do another mockup with mdoc. I chose to export it to html and had to write a simple CSS style sheet to make it look nice, but ultimately mdoc export had some issues and required to apply changes with sed to the output to fix the HTML rendering to not look like a man page misused for something else.
Anyway, I got three mockups of the same Webzine example and decided to use Scribus to export its version as a SVG file and embed it in a html file for allowing web browsers to display it natively.
I asked the Mastodon community (thank you very much to everyone who participated!) which version they liked the most and I got many replies: the mdoc html version was the most preferred by with 41%, while 32% liked the SVG-in-html version and 27% the PDF. Results were very surprising! The version I liked the least was the most preferred, but there were reasons underneath.
The PDF version was not available in web browsers (or at least didn't display natively) and some readers didn't enjoy that. As for the SVG version it didn't work well on mobile phones and both versions didn't work at all in console web clients (links, lynx, w3m). There was also accessibility concerns with the PDF or SVG for screen readers / text-to-speech users and I wanted the Webzine to be available for everyone so both formats were a no-go.
Ultimately, I decided the best way would be to publish the Webzine as HTML if I wanted it to look nice and being accessible on any device for any users. I'm not a huge fan of web and html, but it was the best choice for the readers. From this point, I started working with a few people (still from the same French OpenBSD community) to decide how to make it as HTML, from this moment I wasn't alone anymore in the project.
In the end, the issue is done by writing html "by hand" because it just works and doesn't require extra complexity layer. Simple html is not harder than markdown or LaTeX or weird format because it doesn't require extra tweaks after conversion.
I created a git repository on tildegit.org where I already host some projects so we could work on this project as a team. Requirements and what we wanted to do was getting refined a bit more every day. I designed a simplistic framework in shell that would suits our needs. It wasn't long before we got the framework to generate html pages, some styles changes happened all along the development and I think this will still happen regularly in the near future. We had a nice base to start writing content.
We had to choose a licensing, contributions processes, who is doing what etc... Fun times, I enjoyed this a lot. Our goal was to make a Webzine that would work everywhere, without JS, with a dark mode and still usable on phone or console clients so we regularly checked all of that and reported issues that were getting fixed really quickly.
Let's talk a bit about the website framework. There is a simple hierarchy of directories, one used to write each issue in a dedicated directory, a Makefile to build everything, parts that are common to each generated pages (containing style, html header and footer). Each issue is made from of lot of file starting with a number, so when a page is generated by the concatenation of all the parts parts we can keep the numbers ordering.
It may not be optimized CPU wise, but concatenating parts allow reusing common parts (mainly header and footer) but also working on smaller files: each file of the issues represents a section of it (Quote, Going further, Headlines etc...).
This is a fantastic journey, we are starting to build a solid team for the webzine. Everyone is allowed to contribute. My idea was to give every reader a small slice of the OpenBSD project life every so often and I think we are on good tracks now. I'd like to thanks all the people from the https://openbsd.fr.eu.org/ community who joined me at the early stages to make this project great.
Git repository of the OpenBSD Webzine (if you want to contribute)