But the markdown is not the final product, it still needs to be converted to html using a "static site generator" or some other thing that breaks every couple of years. I had my static site break due to jekyll shenanigans four times in the last few years, and then I decided to move the whole thing to plain html, and replace jekyll and its myriad of unused and fragile dependencies with a three-line shell script that calls "cat". Moreover, markdown is confusing to use for basic stuff like links, images (see the sibling comment), and even enumerated lists [0].
every dialect of markdown has an infuriatingly different way of doing basic things that totally breaks your soul if you use a handful of them at the same time.
EDIT:the example you wrote is not exactly the same thing as mine, is it? It seems like it would convert to H1 and H2 headings and both would be displayed in the body.
> Moreover, markdown is confusing to use for basic stuff like [...] enumerated lists
This is a huge plus for me. Any list with more a dozen of entries and it would be tiresome to have to fix that. Also I can move the lines around freely without weird formatting issues.
Links are okay in my opinion, images are just a bunch of crap. To make an image clickable you'd have to do an image within a link, something like:
[](url)
> every dialect of markdown has an infuriatingly different way of doing basic things that totally breaks your soul if you use a handful of them at the same time.
Completely agree there. Extensions of Markdown are really annoying, but I guess that's what happens when "true" Markdown wasn't updated since 2004.
I agree, and went the html road as well for a time, but I hit roadblocks:
- When I'm in the flow, I really don't want to think about the structure of my text. The ms required for the tag insertion breaks that.
- Unless you very disciplined with your HTML use, it's hard to get back structure from you HTML doc. You have to limit yourself to the subset you would get from markdown.
I'm experimenting with asciidoc now, let's see if it's working.
You can use style="white-space: pre-wrap" on a div, and
deal with the structure later. You'll still be able to use links and images, and the text will not look like a turd, before you get to finishing up the structure. At that point you can remove the pre-wrap style.
> is it mainly about letters and whitespace in sequence like text/plain?
The main inconvenience of writing plain html is the need for <p> tags. They really do break the flow! It would be so much better if a blank line inserted them implicitly. This is indeed the killer feature of markdown and TeX.
I don’t understand these complaints about breakages. Download Hugo. Generate your site. Now use that Hugo binary till the end of eternity. The Hugo project might add, deprecate or remove features but that won’t affect you in any way. Your binary will continue to work till the end of time. (Assuming we still have x86 or ARM computers available)
Never tried hugo, but it surely must be cleaner than jekyll. As a matter of policy I prefer to install tools packaged by my distribution. For debian, the jekyll package went through several "engines" and whatnot along the years, and each change broke the build. For some reason it even needed particular python versions, even if it is written in ruby. A complete nightmare. I honestly lost faith in static site generators after the experience with jekyll. Even if hugo is saner, it is still amenable to breaking as its debian package evolves, and I don't want to deal with that.
If you care about breakage, don't rely on a package manager that will constantly change the package version. Grab the binary from the Github releases page and use it in perpetuity.
> If you care about breakage, don't rely on a package manager that will constantly change the package version.
I do care about breakage of my C, C++ and Fortran programs, and I would be very surprised if updating gcc broke their build (unless using -Werror). Why can't static site generators, which are arguably much simpler than compilers, have the basic property of not breaking sites?
Hugo has no dependency (or rather they'd baked into the single binary), it's just a binary and even runs on my shared hosting. It's the only reason I switched from Jekyll. I had so much trouble with Jekyll breaking it was unbearable. Zola is another option.
That’s quite different: # is <h1>, ## is <h2>. What you were responding to used <title> and <h1>. Markdown is for units of content smaller than a page. It doesn’t provide any means of specifying document metadata or similar.
