@jcabercrombie @anujadas185
Wanted to float a refactor by you, which hopefully won't be too much work and would simplify website maintenance moving forward.
Personally, I've found blogdown to be a bit of a pain point for contribution to the website. Not sure if you have both felt this or if it's just specific to my setup, but I often feel a reluctance to make website changes purely because I find that blogdown is a major annoyance for its inconsistency in how it caches page results (not getting invalidated by style changes, etc).
We already render all of our content into a hugo-ready format. I propose that we just jump directly to hugo and avoid the blogdown intermediate.
What this would mean:
- Working directly in R might be a bit less convenient.
- Instead of
blogdown::build_site() in R, you'd have to run hugo build from the terminal
- When working with an
.Rmd file, you'd have to render it first to .md using your choice of tooling. Then the .md would be used as the page for hugo.
- For non-R pages (that are just pure text without showing R code; which I think is the vast majority of our content), we'd just write the
.md file directly which should cut down on the number of failure points in the deployment process.
blogdown already builds a hugo site, so for any existing content we just need to render it as .md and start omitting the blogdown parts of the build process.
Let me know what you think.
@jcabercrombie @anujadas185
Wanted to float a refactor by you, which hopefully won't be too much work and would simplify website maintenance moving forward.
Personally, I've found blogdown to be a bit of a pain point for contribution to the website. Not sure if you have both felt this or if it's just specific to my setup, but I often feel a reluctance to make website changes purely because I find that blogdown is a major annoyance for its inconsistency in how it caches page results (not getting invalidated by style changes, etc).
We already render all of our content into a hugo-ready format. I propose that we just jump directly to hugo and avoid the blogdown intermediate.
What this would mean:
blogdown::build_site()in R, you'd have to runhugo buildfrom the terminal.Rmdfile, you'd have to render it first to.mdusing your choice of tooling. Then the.mdwould be used as the page for hugo..mdfile directly which should cut down on the number of failure points in the deployment process.blogdownalready builds a hugo site, so for any existing content we just need to render it as.mdand start omitting the blogdown parts of the build process.Let me know what you think.