Pages made of images
Every image is written by generate-assets.sh and none of
them is in the repository. Each file is unique, so nothing here can
be served from a cache after the first request.
By count
- 50 images — around 2 MB of page weight
- 250 images — half of
crawler.requests.max_per_page - 1,000 images — twice the request budget, so the report has to say how much it left out
By behaviour
- Lazy loaded — a thousand images below the fold, most of which should never be fetched
- srcset and picture — one image, five candidate files, one request
- Broken sources — a hundred images pointing at files that do not exist
- No alt attributes — a hundred images nobody named
- Data URLs — a hundred images that cost no request at all
- CSS backgrounds — a hundred images no img element accounts for
- One very large image — a single file of about 8 MB