Why static website generators suck so much - or why it seems to be impossible to get working software without writing them myself. None of all existing generators are coming close to the performance of LaTeX2HTML or stuff like mandoc.db seems to be a lot more suitable, faster and reliable.

Here are my points:

  • Jekyll is fantastically simple, however it is getting really slow once we created a few posts and pages
  • The time it takes to rebuild the whole website can vary quite a lot (tested under lab conditions)
  • Hugo is much fast, however Hugo’s syntax sucks and it is almost impossible to import a Jekyll project once we use a lot of html and css inside markdown
  • Pretty much all static site generators rely on some flavor of markdown
  • HTML, CSS and JavaScript still suck like they did 15 years ago - perhaps web assembly makes certain things a bit better but the frontend will rely on html and css hence I doubt it

What pisses me off with Jekyll:

  • it is slow and somehow there are ways that pages are not going to be build - I just haven’t figured out why. It affects ~ 30 pages, currently.
  • certain html+css structures seem to break jekyll - this is sad considering the severe limitations of markdown
  • markdown itself is limited so much that slightly many of the more technical things are not possible
  • certain ruby gems are broken and according to some bug tracker it seems like either no one cares or some of it is deprecated

Possible solutions:

  • rewrite Jekyll in Rust (FORTRAN and Julia are less optimal for handling text IMHO)
  • use another static site generate that requires less adoption than hugo
  • going back to (almost) unformatted HTML - since 99 % of my readers are extremely deep into computer science they probably don’t mind

I haven’t made up my mind yet. I upload blog posts less frequently until I solved this. Hence, only blog posts will be uploaded that use standard markdown (liquid). Everything else is prepared but will be uploaded once I made up my mind (if you find pages that show titles only then you ran into one of those pages). The downside of this is that you’ll have to wait a few more weeks for really technical blog posts that have interactive elements and cover more than a bit source code and some text around it.