okay I'm officially that kind of person

If power is plentiful: make devserver

If every joule has to be accounted for:

• python -m http.server 8000
• make html to rebuild when wanted

cd comment-server
poetry install
CONTENT_DIR=../output/ COMMENT_DIR=../comments/ poetry run gunicorn -b 0.0.0.0:5000 app:app

always submits to /newcomment, with the article slug in its url-encoded payload (via an <input type="hidden" />)

dockerized flask server, running with gunicorn on my 4-year-uptime vps, port 5000

build images

cd comment-server
docker build -t jpcapurro/comment-server-base -f base.Dockerfile .
docker build -t jpcapurro/comment-server .

push image:

docker save comment-server:latest |xz|pv | ssh selfhost docker load

run it:

ssh selfhost 'docker run -v $HOME/comments:/home/app/comments -v /var/www/blog.capu.tech:/home/app/output -p 5000:5000 -d --restart always comment-server:latest'

writes comments to a directory on the host, one file per valid article slug.

What I do is pull the files with a good ol' rsync -av selfhost:comments .

And then git add -p comments to review them before rebuilding

good ol' nginx running on my 4-year-uptime vps.

serves static files for the blog itself, which are updated with make rsync_upload

requests for /newcomment are reverse-proxied to port 5000