search

ReproNim / repronim.github.io

http://www.repronim.org/
GNU General Public License v3.0
3 stars 10 forks source link

Investigate Website Stack options #354

Open asmacdo opened 2 months ago

asmacdo commented 2 months ago

Lets put together a pros/cons list of various options.

To name a few:

asmacdo commented 2 months ago

From @yarikoptic For web folks who might still be choosing website build helper. Michael Hanke seems to quite like hugo (https://gohugo.io/), and its feature of "taxonomies" where from metadata for the pages, hugo could build various "listings" and also provide "permalinks" to the term values. See more about taxonomies https://gohugo.io/content-management/taxonomies/ and a basic example at https://trr379.de/contributors/ . In the scope of repronim with various people participating in various projects and all the "semantic web" notion, sounds like a reasonable tool/approach :wink:

satra commented 2 months ago

i would vote for hugo instead of jekyll today. or just a jupyter-book or quarto or some such which can bring some of computation elements into our site as well.

asmacdo commented 2 months ago

I've also used hugo to build operatorframework.io and for the most part I thought it was pretty nice. In that case, I chose hugo to align with the rest of the kubernetes community, which uses Hugo heavily (including kubernetes.io itself)

Hugo's integration with netlify was very good (free hosting for open source projects) and included PR previews "out of the box".

The challenging part of working with Hugo for my previous use case was the theme stuff. In particular, extending the docsy theme was problematic, though the theme was still pretty new all those years ago.

asmacdo commented 2 months ago

FWIW I included sphinx because this is a python-heavy community, but I suspect its not the right choice for us. Its been a while but I found the rst-style was more cumbersome than the markdown used by Hugo or Jekyll. I suspect our community would prefer using very similar syntax to GitHub rather than learning rst.

asmacdo commented 2 months ago

i would vote for hugo instead of jekyll today. or just a jupyter-book or quarto or some such which can bring some of computation elements into our site as well.

@satra can you elaborate what you have in mind for computational elements? (I imagine live jupyter notebooks would be really cool) IMO, this is a strong reason to continue focus on the high level requirements prior to making the stack choices.

satra commented 2 months ago

here are some examples of data books:

it's all markdown based and also notebooks that can be embedded.

in terms of requirements, being able to execute a notebook might allow us to do some things like process data to generate figures of stats or other things from courses, usage, etc.,. i would put it as a nice to have, but not necessary requirement.

yarikoptic commented 2 months ago

in terms of requirements, being able to execute a notebook might allow us to do some things like process data to generate figures of stats or other things from courses, usage, etc.,. i would put it as a nice to have, but not necessary requirement.

for mkdocs we are now about to include them via mkdocs-jupyter in https://github.com/dandi/handbook/pull/145 . And if need arises we could always add them up somehow e.g. as in https://medium.com/@romankurnovskii/how-to-add-jupyter-cuaderno-in-hugo-static-site-67cfdb42653f . And for interactive ones, we better run a hub anyways ;-)