Please read Philsophy and goals for background first.
We'd also like to link to an Architecture document about why things are this way, but it needs to be written first.
- Download the current state file from S3.
- Load a directory of JPEG files.
- Create thumbnails of all required sizes.
- Compress these thumbnails (not the original).
- Strip metadata.
- Generate a UUID, put it in the (local copy of the) state file.
- Upload thumbnails to S3, record in state file.
- Upload the new state file to S3.
Generate and deploy
- Download the state file from S3.
- Generate an index page with every album.
- Generate an album page with every image from that album.
- Generate an image page for every image in the album.
- Put these in a directory.
- Upload this directory to a webserver.
- Call a deploy hook.