#Artistic new tab

When browser.newtab.art.enabled is on, a new tab shows a random public-domain painting filling the window, with the start page's shortcuts and the address bar legible over it, a discreet caption (title, artist, date, museum, a link to the record) and two small buttons (Another painting, Open the record). It follows Google Arts & Culture's "art new tab" idea, but every piece runs natively in Wren: the engine renders the page, the shell downloads and caches the images. The option is off by default.

#When the network is touched

Never at start-up, never while the option is off, and never while the loader is held offline. The first time an art new tab is actually shown in a session, a background task starts; from then on it keeps a sliding window of about browser.newtab.art.window paintings ready, downloading a fresh one whenever a shown painting is consumed. The task stops after browser.newtab.art.idle_minutes with no art new tab shown and resumes at the next one.

The only requests made are the source's listing API and the image files themselves. Nothing about the user is sent: no identifier, no third-party referrer, an honest User-Agent (Wren/<version>). The images and their metadata live only in the profile's temporary folder (<profile>/tmp/newtab-art), one JSON file of metadata beside each image, and are removed with Clear browsing data (the "cached responses" kind, for all sites).

#Sources

browser.newtab.art.sources is a comma list tried in turn (replaceable):

• cleveland — Cleveland Museum of Art Open Access. CC0 paintings, no key. A random slice of the ~3,900 CC0 paintings is listed; the larger images.print rendition is used, falling back to images.web (~900 px).
• commons — Wikimedia Commons. The Google Art Project works by artist and Featured pictures of paintings categories, listed with generator=categorymembers and prop=imageinfo (iiurlwidth for a bounded width, extmetadata for artist, title and licence). Only freely licensed files are kept (public domain, CC0, CC-BY / CC-BY-SA); non-commercial, no-derivatives and fair-use files are refused. The attribution is shown in the caption.

Google Arts & Culture itself has no public API and is not used. The Met (search returns 410) and the Art Institute of Chicago (Cloudflare challenge) do not answer a client without a browser reliably and are not used either.

#Courtesy and validation

The downloader makes one request at a time per host and backs off exponentially on 429 / 403 / network error (browser.newtab.art.backoff_base_ms doubling up to browser.newtab.art.backoff_max_ms). Every answer is validated before it is kept: an image/* MIME type, a size below browser.newtab.art.max_image_bytes, a known image signature, a probe decode with Wren's own image decoder, a sane aspect ratio (between 1:3 and 3:1) and a minimum width (browser.newtab.art.min_width_px, 1280 px by default) for full-screen use. Files beyond browser.newtab.art.max_cache_mb are purged oldest first.

#"Smart"

The painting chosen for a new tab suits the window's orientation (landscape for a wide window, portrait for a tall one); a painting of the other orientation is shown whole over a blurred fill of itself rather than cropped to a sliver. The last browser.newtab.art.window picks are remembered so the same painting does not repeat at once, and the next one is pre-fetched while the current is on screen. Nothing downloads on a limited or offline connection.

#Preferences

#Trying it

1. Open about:config, set browser.newtab.art.enabled to true.
2. Open a new tab (Ctrl+T). The first one may be the plain start page or the fallback gradient while the first paintings download; open a few more tabs and they fill with paintings.
3. Another painting (or opening about:home?art=next) shows the next one; Open the record opens the museum's page for the painting.
4. Clear browsing data (about:sites, "cached responses" for all sites) empties the cache; the task refills it.

Edit this page on GitHub