Ivy – A static website generator built in Python

I think if people dedicated time to creating one for their own use it can be done (I am decent at Python and I would like to think that I can too) but at the same time I don't want to much around with HTML and particularly CSS.

I use static generators purely by available themes because that's the part that I hate the most.

I'm one of those people who falls into that group -- I can, and have, used a static site generator, but have nearly zero programming knowledge so couldn't make one of my own. The biggest issue I have now is choosing which one to use. I've mostly used Hugo, with good results, but always have that slight niggle wondering if there's something that would work a bit better for me.

>Is there an in-between type of user technical enough to use a static site generator but not able to write their own?

A lot of people who want pretty websites without writing or learning a lick of frontend technologies.

>Is there an in-between type of user technical enough to use a static site generator but not able to write their own? Or maybe the proliferation is only because they are both easy and fun to create?

Not sure if you mean to ask why people don't create their own static sites (without generators), or did you mean dynamic sites using PHP/Ruby/Python?

For the argument over dynamic sites:

Yes - the type that does not want to worry about security vulnerabilities.

My blog is made via a static site generator. I updated infrequently (once every so many months). I want to have it up and running for years without my intervention (i.e. maintenance).

I once had a Wordpress blog and treated it that way. It was hacked.

Then I built my own in Django. Then at some time it went down because my service provider updated Python libraries, etc.

The funny thing is: Using a static site generator is no more work, and has no fewer advantages. Why should I use a dynamic site or build my own?

For the argument against static sites, well then you'd have to maintain lots of links manually. A SSG gives you a lot of that for free. And you can use templates.

I fall into the category of a technical user who could build their own, however I use existing ones as:

- It's faster and I already have more fun projects than my time allows.

- I don't particularly enjoy writing HTML/CSS (though the rest of the project would be fun).

- I'm happier using an existing template rather than making my own. I know I'll be too critical of the design if I do it myself.

>It looks nice, but I guess I still[1] don’t understand the proliferation of simple static site generators. It’s pretty easy (and fun!) to build your own that works exactly as you want it to.

I don't think that's true. It looks deceptively easy up front but the subtleties involved actually make it pretty hard. I've tried hugo, jekyll and pelican and they've all pissed me off for one reason or another. A common issue involves one of the 'template' themes I found on their template theme libraries not working on the "latest" version of the generator.

Ivy seems to have "solved" this problem by having almost no themes. This is not exactly the solution I was thinking of...

There's still a gap in the market here I think, and there will continue to be a proliferation (like how there was with bad javascript toolsets until jquery 'won' ~2007-8) until somebody makes an acceptably 'good' one or fixes an existing one until it obviously stands head and shoulders above the rest.

If you can make a well designed static site generator with lots of nice themes (or the potential for that) that doesn't fail horribly when I try to use it in a normal fashion I'd switch to it in a heartbeat and tell all of my friends. I'm using hugo now, but I'm not super happy about it: last problem with that being that I couldn't get it to competently handle breadcrumb navigation.

Learning how to program, especially when one doesn't come about it through self-selected autodidactism, can often seem to be an insurmountable challenge to people who don't know how.

This applies to even extremely technical, highly intelligent people. Often they recognize that there's an extraordinary time component to learning, and they optimize around that in other ways.

Learning to program is a weird thing because once you learn the mental processes of how to do it, it comes fairly easily. But the idea of "I'd like a tool that makes websites for me" and then knowing how to divide that problem down to the right mental abstraction model that perfectly (or near perfectly fits) some programming language and technical environment is a very difficult learning curve.

It's much simpler then to present small technical pieces that are well constrained in scope and definition and have tooling somebody else built do all the rest of the heavy lifting.

Static site generators are one of those things that are relatively easy, even for newish programmers, to build, but exist right on the other side of that learning curve for non-programmers.

I've a site deployed using Lektor[0] for a slightly technically inclined user, who gets a UI to edit his site and when ready to publish, just click the button.

[0] https://www.getlektor.com/

> I guess I still[1] don’t understand the proliferation of simple static site generators

> It’s pretty easy (and fun!) to build your own

Seems like you answered your own question. :)

Static site generators are more than suitable for non-tech users - they used to be called Macromedia Dreamweaver, Microsoft FrontPage; etc.

I'd very much like to see fullon wysiwyg static generators again; it could replace half, if not more, of the WordPress sites around.

>It looks nice, but I guess I still don’t understand the proliferation of simple static site generators. It’s pretty easy (and fun!) to build your own that works exactly as you want it to.

Notice how these two sentences are contradictory -- or rather how the second explains what the first supposedly can't understand.

If it's "pretty easy and fun" to build X, then that would inevitably lead to a "proliferation of" X projects.

It just takes a person that built one to then share their implementation on GitHub to increase that proliferation, and that's a very simple additional step.

And it just takes any person who understands "opportunity cost" to want to adopt an existing static site generator rather than build their own.

Both kinds are in abundance.

I agree it can be fun. And lots of us probably have written such systems ourselves. However, even for those of us who could write one, sometimes you just want to use a tool, not write it and worry about bugs and new features constantly. A good pluggable static site generator can potentially give you a lot more featues than something you write yourself.

Sure, I could write my own. But, for god's sake, why? My time is too precious to spend on all things that I could do on my own.

I find the features of Jekyll important; I use it for client projects and mock up designs. While I don't like everything about how it's structured and I've written my own for fun, it'd be a lot of work on my part to create and maintain something equivalent.

We're helping bridge the gap for non-technical users with our Jekyll and Hugo CMS (https://forestry.io). There are many other options too, just check out https://headlesscms.org.

It's an exciting time to be a web developer. We have many static site generators (like this one) to help build super performant sites, we can host them in the cloud for a negligible cost, and we can keep everything in Git.

Pelican is great, I find it very flexible thanks to its many plugins. I wished it was maintained a bit more though, there are a lot of PRs waiting for reviews[0].

[0] https://github.com/getpelican/pelican-plugins

I didn’t care for how it pushes a web server and absolute urls on you. Seems it could still learn a lot from Sphinx.

As the sibling comment mentions I sent a pull request to fix some broken css and they requested I fix their test suite first. No thanks, moving on.

Not OP, but I'm tempted to switch. Because I had an issue with Pelican where I couldn't get year/month archives to work sanely (Pelican does year/month/day which is too fine for me). So I had to write my own generator anyway, and plumb that into Pelican. It wasn't that easy to do, although I do think Pelican is fairly hackable, if anything else was more easily hackable/extensible, I'd use it. Plus not having Python 2 legacy stuff in the codebase is a plus when I'm grokking it. I also find the Pelican config overburdened [0], but understand why it came to be that way.

I'm giving it a go this weekend, let you know if it worked out for me.

[0] http://docs.getpelican.com/en/stable/settings.html

Thanks for the clarification. I was looking everywhere for a price tag but couldn’t find it. Was excited to see another dev charge for their work! Maybe the title of this post should be changed so it’s less confusing?

You can get some decent mileage out of a framework like Bootstrap[0] or Foundation[1] without diving too deeply into it, but it really depends on what your tastes are. I'm not sure about the WYSIWYG options out there though.

[0] https://getbootstrap.com/

[1] https://foundation.zurb.com/

* Avoid custom fonts. You probably don't need a custom font. Just use system fonts. Using built-in fonts will allow your site to load faster.

* Getting a "nice layout/theme" is a design problem. I'd suggest following someone's style guide. I'm fond of the U.S. Web Design System [0] since it's lightweight and accessible. Disagree with their suggestion to use a custom font.

* Keep the layout as simple as possible, with a focus on content. You can go incredibly far with semantic HTML and a smattering of CSS. I don't think it requires a deep-dive.

[0] https://designsystem.digital.gov/

I know this is controversial but the default html layout is what most people need. You can use markdown or org mode etc. to generate it. One improvement may be centering the content on bigger displays, which I do for my blog with two @media rules as follows: https://www.gkayaalp.com/style.css

Site was done in php with peak user load being around 200 simultaneously users. I had git on the webroot and a chron job that dumped the sql, and pushed the whole thing every night.

The fear-mongering about public domain seems to drift further and further from reality. Is SQLite "impossible" to use globally? Its public domain dedication is even more terse than the unlicense. If it really does become a practical issue for Germans, maybe the real problem is the German legal system.

In more-or-less the same niche as Gatsby there is my Sitio[1], which also uses React (so kinda isomorphic pages), main differences being:

- No GraphQL stuff, but instead you'll use imperative JS code to tell the library which pages you want to build, with data from anywhere (files, CMSs, hosted contents elsewhere etc.) you can fetch using the same JS code, transform with any library etc.

- No Webpack, no Webpack transforms. I suffered when I tried to use Gatsby because of the gigantic dependencies on Webpack. I guess it's changed nowadays, but still I'm hurt and won't go back there. Here we're proud to use Browserify and go with more modular architecture.

[1]: https://github.com/fiatjaf/sitio

For me, Ivy is a Emacs package (interactive list narrowing/selection). https://youtu.be/VvnJQpTFVDc

Depends on width of the monitor, of course, but one reason would that it is easier for reader to follow lines if they are not too wide. The same reason is behind multicolumn layout in newspapers.

The best length for fast reading is not more than 60-70 characters per line. That's why you have columns in newspapers. I have a 4k 26 inch monitor. Full width texts are harder to read, I have to shrink my browser to make the text not so wide.

The main advantage of a static site generator is it allows you to just write the contents of each page and the generator will insert that into a page template. It means that, for example, a site-wide navbar only needs to be written once rather than added to every page manually.

Suppose for instance you wanted a static blog. Blog engines like wordpress (or django or flask for python setups) generate parts of the site on the fly as users visit. For instance, you might generate the most recent 10 posts in some list somewhere. This requires some (often not much, but some) additional time from the server as these are generated on the fly.

A static website generator like this allows you to make a blog (for example) and update content like most-recent-10-post lists (also for example) only when you actually add a new post. Thus no additional time is required on the fly.

I might add that there is an extension for flask (which I normally think of as dynamic) called Frozen-Flask which also allows a flask-built site to become a static site. This probably exists for django as well, but I'm less familiar with django than flask.

In HTML you cannot reuse code or split the code into multiple files without using copy and paste, iframes, or Javascript.

Software like this allows you to create templates and generate multiple pages with different content.

This can be achieved with server-side scripting languages. However, it is slower than static HTML. So if the content doesn’t change often, generating static HTML might be the eay to go.