Jan Miksovsky’s BlogArchive AboutContact


Sample influencer lifestyle blog for Henry David Thoreau

To demonstrate that Origami is a good language for building #smallweb / #indieweb blogs, I built a fun sample blog reenvisioning Henry David Thoreau has a modern influencer with a lifestyle blog about off-grid living.

Blog post titled "Beans" with text adapted from Thoreau's "Walden"

Creating a site in Origami is completely different than creating a site in any other tool I know of.

Origami isn’t a blog engine or framework, just like Microsoft Excel isn’t an invoicing or expense reporting framework. Excel is a general tool that transforms and aggregates numbers and text in tables. Origami is a general tool that transforms and aggregates data and content in trees.

So this project isn’t configuring a blog tool — it’s defining what a blog is from scratch. The source is endlessly malleable, and you can readily change what gets produced to achieve a wide variety of results.

To build in Origami, think about the starting tree of content you’ll write or gather by hand, and the final tree of resources in your running site. Step by step, you transform the former into the latter.

For a blog you might start with, say, a markdown folder containing markdown posts, an images folder for photos, and an assets folder for stylesheets.

Tree diagram of staring point with assets, images, and markdown folders

The final resource tree comprises the images and assets as is, plus generated pages for individual posts, list pages, and feeds.

Tree diagram of site resources showing generated pages, posts, and feed areas

To generate individual pages you can use a template language. Origami has a nice one built in; you can use others.

But the real magic is writing formulas to process a pile of content at once, like a pipeline that transforms markdown to HTML, calculates data, and sorts the posts to prepare them for rendering.

Origami pipeline transforming a markdown folder into data

Or a compact definition of the entire public portion of the site:

Origami program that defines the publicly-visible resources for a blog site

This Origami program generates the full tree of the resources for your blog, which you can browse immediately. The Origami runtime only does the work to generate a page when you ask for it.

The same Origami program can also produce a complete build folder with all your site files. Deploy those on a static web server or have a service do a build whenever you update your project.

Full source

Sample Origami site for an outdoor trekking company

Each month this year I’m trying to post a sample website written in Origami, a declarative programming language at the level of #HTML and #CSS for defining websites. This month’s sample is Aventour Expeditions, a site for an outdoor travel company.

It’s easy to have Origami call other template languages, so for this sample I used the Handlebars template language to turn markup and data into HTML.

Home page for an outdoor trekking company with tagline "Start your adventure" Page for a trek to the Wadi Rum desert in Jordan

It took just a few lines of Origami code to define the structure of the site and indicate which Handlebars template should be used to create which pages. For this outdoor travel example, Origami makes it very, very easy to:

Origami site definition for the trekking company site Tree diagram of the trekking company site

Origami also made it easy to cross-link the data for each trek with related treks. That allows the page for one trek to show cards links to related treks.

Page showing related treks to mountain destinations in Canada and Nepal

Aside: modern HTML and CSS are soooo much better than the past. I based the trekking site on a WordPress template whose design I liked but whose HTML/CSS was ridiculously complex and burdened with tons of JavaScript for trivial things.

Rewriting slashed the size of the pages. For the home page:

Before: 42k HTML, 600k CSS, 1580k JS

After: 13k HTML, 12k CSS, 0k JS!

Using Handlebars templates in the shell

It’s useful to be able to apply templates written in a language like #Handlebars to things in the shell.

The Origami #CLI lets you invoke JavaScript functions defined in .js files, but you can now also identify a handler for any file extension — like OS app file associations, but for a CLI. https://weborigami.org/language/filetypes.html#custom-file-types

So a handler can load a .hbs file as a function that applies a Handlebars template, then apply that in the command line.

Origami can resolve pretty much anything — JSON/YAML files, files with front matter, entire folder trees, a ZIP file, whatever — to an object your Handlebars template can operate on.

You can even write a one-liner that fetches data from a server and applies a template to it.

Applying a Handlebars template to a file system folder lets you, e.g., generate an HTML index page with links to all the HTML pages in the folder.

If one Handlebars template references another template (a “partial”), Origami resolves that reference for you. If you reference bold, it will look for a template named bold.hbs and use that.

This lets you decompose templates for complex results without the usual overhead of manually loading those templates and passing them as Handlebars configuration.

Origami itself is a powerful template language, but you might prefer Handlebars or some other template language. You can use Origami to define the overall structure of your site, then use Handlebars/etc. to turn data from anything into HTML.

It was easy to rewrite the Origami “About Us” sample site using Handlebars.

All City Someday: a photo blog backed by Google Drive

Suppose you want a photo blog that’s not on Tumblr or Instagram. Maybe you don’t like being sold, or want to draw outside the lines, or want photos as part of your own site. What would a #smallweb alternative look like that let you post a photo and caption from your phone?

I made a proof-of-concept street art photo blog that supports this on-phone flow: 1) snap a photo and set a caption, 2) save the photo to Google Drive via Files, 3) trigger a rebuild via the browser, 4) see the new photo on your own site.

View source

Google Drive is hardly #indieweb, but look beyond “corporate = bad” and consider the nature of the relationship. They provide a commodity service: you pay them, they store your files. Those files are a medium of exchange. If (when?) Google screws around, you can easily move the files to a different storage provider. Switching costs for hosted platforms like Tumblr or Instagram are far higher.

All storage services provide a proprietary API to enable unique features and faster service… but also to force you to write to that API and make your own code a switching cost. But most of what you want from storage is just an easy way to: 1) list directories, 2) read a whole file, and (maybe) 3) save a whole file.

The tiny AsyncTree interface does exactly those things.

interface AsyncTree {
  get(key: any): Promise<any>;
  keys(): Promise<IterableIterator<any>>;
  set?(key: any, value: any): Promise<this>;

This interface lets small storage drivers wrap the proprietary APIs; switching providers is just a matter of switching drivers.

The Origami language can use such a driver to read images out of Google Drive and create a static, deployable site. Origami templates also understand that interface, so you can make a template that directly maps a folder’s photos to HTML <img> tags with dates and captions. 🎉

One significant hurdle: obtaining a Google API credential file. OMG figuring out any kind of auth on Google, Azure, or AWS is like wandering through a maze of twisty little passages, all alike. It’s also theoretically possible to use Google Cloud Pub/Sub to automatically trigger a site rebuild whenever a photo is added to Google Drive, but dang that looks complicated. Maybe some other day.

Standards like Exif (for photo captions and dates) make amazing things possible. They’re there for everyone to use — if a corp isn’t trying to cage you in their silo.

The code for this photo blog is small: HTML + CSS + about 40 lines of Origami code. That includes defining RSS and JSON feeds. 😃

If I ever needed to switch storage providers, I’d just have to update the 3 lines of code that authenticate with Google and get a reference to the Drive folder. Everything else would stay the same.

Abstracting away storage doesn’t just give you an interesting way to build a site, it also opens up general-purpose terminal access so you can:

Design notes:

  1. It was fun to play with #CSS blur, text-shadow, and -webkit-text-stroke for a graffiti heading that’s just plain text! May also be the first time I’ve ever used magenta.
  2. I was able to auto-size the header width to be roughly as wide as the photos by playing with values for font-size: clamp(…).
  3. The fonts are so different from the stock fonts that any font swap was really jarring, so I forced early loading of the small font files and used font-display: block.

I made an ebook about my hike across Japan

I’ve made a little ebook about my hike across Japan last year.

In the summer of 2023, I walked for a month from the Pacific Coast to the Sea of Japan. I crossed steep mountains including Mt. Fuji, portions of the Southern Alps, the Yatsugatake range, the Northern Alps, and the Shio no Michi Trail. In between, I walked through cities, towns, and rural areas. The hike was a substantial challenge but also a fun adventure.

Book cover showing a grassy meadow and an elevated walking path

Creating an ebook is not difficult. Will Crichton wrote a great article advocating for using the EPUB format as a more portable and responsive alternative to PDFs.

Before reading it I hadn’t appreciated that EPUB files are just ZIP files with a particular folder structure and some book data. In retrospect that makes sense, but for me that single insight demystified the entire prospect of creating an ebook. It’s just a small bundle of files, mostly HTML, and some metadata.

You can use apps to make an ebook, but I wanted to build one more directly. Since the Origami website definition language is great at processing trees of content (like a folder of text and images) into new forms, I realized it could easily generate an ebook in EPUB format.

It wasn’t hard to write an Origami program to generate files in the required folder structure, and then it was just a matter of zipping it up.

Which left the question: what material could I turn into a book?

JamesG posted a good list of ideas for things people can do on their website. Facundo Olano expanded on that to suggest adding EPUB downloads for your blog or subsets of it. Their posts made me realize my own hike posts might make an interesting sample book.

During the hike I’d kept a travel diary in markdown with photos, so it was fairly easy to add a cover image, an introduction, and some metadata. Voilà: a book!

It may seem silly, but packaging something as a “book” feels weightier. Simply seeing the cover art in my ebook library alongside purchased novels was thrilling, as was opening up the book and paging through it. It felt like that day in third grade when a teacher had us bind illustrated stories into books and the school librarian put them on a library shelf. 📚

Try making an ebook — about anything — and you’ll feel the same thrill.

Small Web Build of a little Cat Prints online store

For this month’s #SmallWebBuild, I made a little sample store for selling art prints.

The site’s structure is defined in a single file in the Origami site design language that turns the sample artwork images and data into a gallery home page and a separate page for each print.

I built this to try out Snipcart, a web service for adding a shopping cart to a static site. They seem to charge lower fees for this service than the larger site hosting platforms, and it was trivial to add the shopping cart with a few HTML attributes. I love that their approach leaves you in control of the rest of your site. That said, I haven’t signed up to use Snipcart to actually complete financial transactions; if you’re interested in Snipcart, investigate it carefully.

I also used this store site to experiment with the new web View Transitions API. I was able to create a zoom effect that works across page navigations (!) which is completely nuts. Caveats: currently in Chrome/Edge/Opera only; it was hard to figure out how to do what I wanted; Chrome sometimes shows a white flash before the zoom-out animation. Still, I’m excited to see these effects coming to simple multi-page sites.

Internet archives are amazing and under-appreciated.

I searched a #Usenet archive for a program I’d written decades ago in high school: an interpreter for the Karel programming language for use in intro CS classes. Because wonderful people have maintained those archives for many years, I was able to find my program 40 years after I wrote it. https://www.usenetarchives.com/view.php?id=net.sources&mid=PDI5OUBzcG9jay5VVUNQPg

It’s hard to describe what it’s like to see that #C code again now. (Although I’m frankly appalled that teenage me didn’t put curly braces around one-line if clauses and for loops.)

I downloaded the source and was delighted that the ancient #Unix shell script to extract the code still runs just fine in bash. (It would take much more time to actually get the program itself to work again.) I posted the source on GitHub: https://github.com/JanMiksovsky/karel.

40 years from now, which is more likely to still exist: the USENET archive version, or the GitHub version? 🤔 I’m betting the former.

When considering modern web development, I keep two images in mind.

Gleaming Tesla gigafactory with robots and no people

This factory image captures how many devs approach creating websites: constructing an enormous, gleaming, automated system that can produce complex artifacts on a massive scale.

A potter using their hands to sculpt a clay pot on a wheel

In contrast, an individual artisan working in a studio directly creates something simple but useful using their own hands and understandable tools. To me this second image captures the #smallweb/#indieweb spirit.

Factories aren’t always bad and artisans aren’t always good, but many devs can’t conceive of building anything without first making a big factory. “You’re creating that pot by hand?? Why, a gleaming pot factory could make a million pots a day!”

That’s ridiculous. A factory — even one magically built for you — can easily become a monstrosity you don’t understand but have to maintain. In many cases simple tools produce better results than a factory. And some of us like making things by hand!

The Origami language gives web creators functions for making parts of web sites in the same way a spreadsheet offers functions for crunching numbers: give a function some input, get something useful back.

Case in point: every so often I want a web page to show a screenshot of another web page. Puppeteer is great for that but cumbersome to set up, so I made screenshot functions: give them HTML or a URL, get back an image.

I used this to make a new Examples page on the Origami docs site. A build script uses the screenshot function to take pictures of each sample URL, producing a virtual folder of images that get saved as real images for deployment. This will make it easy to add new samples or update the screenshots as the sample sites evolve.

Small Web Build video of sample Cherokee Myths site

I enjoy videos of people showing off small homes they’ve built or van conversions they’ve done. The person shows off their craftwork, describes goals and trade-offs, recommends tools/components/techniques, and shares lessons learned. It makes building something feel approachable and can inspire people to try it themselves.

I’d love to see #smallweb/#indieweb creators do the same with #SmallWebBuild videos. Here’s my first.

This walks you through the Cherokee Myths site I described previously that uses Origami to create a static site with a generated table of contents, plus a full-text search feature using Pagefind by Liam Bigelow.

Using small home and van build videos as models, the following could make for good #SmallWebBuild videos:

Cherokee Myths sample Origami site with a table of contents and full-text search

I built a sample Cherokee Myths site to explore how easy it is to create a static site in Origami that includes a generated table of contents and full-text search. I’m happy with how the site turned out.

Cherokee Myths site home page with a table of contents listing myths grouped by topic

As described in the site source, the site’s overall structure is defined in a single concise Origami file that orchestrates the creation of the table of contents on the home page, the generation of search indexes, and the processing of the tree of markdown content into HTML.

Diagram of navigational structure of the Cherokee Myths site

The tree of stories is flattened by a template to create the table of contents, and that same tree is also fed to a great tool called Pagefind to generate search indexes to allow full-text search on the static site.

Search page showing that "hummingbird" produces two hits

Using public domain content for this sample was way more interesting than lorem ipsum text, and the Cherokee myths have a distinctly different mood than Greek myths or Aesop’s fables. The tale of the The Race Between the Crane and the Hummingbird did not end the way I’d expected.

Cherokee Myth of the Race Between the Crane and the Hummingbird

This sample also gave me a chance to place with two recent CSS additions: text-wrap: balance so that long headings end up with lines roughly equal in length (instead of a long line followed by an orphaned word), and initial-letter to add a drop cap at the start of the story text.

Update: I made a video tour of this site.

Origami language as a complement to HTML/CSS/JS

Writing about the incompleteness of HTML/CSS/JS has helped me reframe my Origami language as a way to complement those native web languages.

People with some experience with HTML and CSS (but maybe not JS) have a world of ideas to share but may struggle with those missing platform pieces. Origami is a small language to round out the standard ones so people can make even cooler sites, especially small and independent sites.

The Origami language is small and focused on common web tasks. I take inspiration from the core W3C principle of the Rule of Least Power: choose the least powerful programming language for a given purpose.

Web diagrams look deceptively complete

I see so many web platform diagrams that represent HTML/CSS/JS in a tidy, complete arrangement that suggests those are everything you need to know — when in reality those only let you define what happens in individual pages. A more realistic diagram would be incomplete! You’re going to need to choose and master additional technologies to create a coherent, functional site.

Two bubble diagrams: Left shows HTML, CSS, and JavaScript; Right adds bubbles with question marks for site structure, server, and HTML reuse

Even the simplest possible static site requires site structure (organizing the files into a folder hierarchy) and a server that can respond to web requests with those files. A non-trivial site generally also requires some way to reuse HTML across pages. Where is all that represented in the first diagram?

JavaScript developers will happily explain that their preferred language can be used for all those purposes via Node, so for them the first diagram is actually complete. That’s great for them! (I use Node too.)

But many people don’t want to learn JavaScript or find it too difficult, and for basic sites it’s massive overkill. Moreover, the diagram on the left suggests you can’t build a page without JavaScript, but I think JavaScript on the front end should be used sparingly.

Web authoring tasks you’d think should be easy but are surprisingly hard

I wish it were possible to create more complex sites just with HTML and CSS — but any aspiring web author learning to create a site that way quickly finds some very common task that’s impossible in plain HTML and CSS.

Author: How do I make my nav bars do that thing all nav bars do: where it shows the link to the current page differently so my user knows where they are?
Web: Ah, yes, you want to do that! But you cannot. The browser certainly has all that information necessary to do that for you — you might have even been intrigued by the :local-link CSS property. That does just what you want! But no browser supports it.

Author: I want an Articles index page that links to all the articles I wrote in the /articles folder, and I want my Gallery page to show all the photos I have in the /photos folder.
Web: Sorry, HTML just deals with what goes on a page. We have heard about these “folders” you’re talking about. They sound very popular! Not our thing, though.

Author: I want my blog post pages to have links to the next post and the previous post.
Web: HTML defines what goes in a single page; there’s no way to represent a set of pages or interrelationships like ordering, much less reflect those relationships in a page. Think of HTML as an exciting “Build your own adventure”: you can do whatever you want — by hand!

Author: I want all the pages in my Products area to have the same layout, with a product name, photo, and description. I don’t want to have to repeat all the HTML required to do that.
Web: That sounds very useful! Again, though, that’s not really HTML’s thing.

Author: I want to create my navigation bar in one place, then reuse it on every page.
Web: Sorry, everything needs to be directly in the HTML! Well, yes, the browser can load CSS from other files… and JavaScript from other files… and SVG images from other files… just not HTML. Have you tried changing careers and learning JavaScript?

I understand at a technical level why these limitations exist. Working in web standards gave me an appreciation of the political difficulty of solving such problems even where solutions are possible. But none of that makes a new web author feel any better. Watching someone struggle to create a simple site makes clear that some very common authoring tasks are still disappointingly hard.