a list can keep out of the file of its page

[pfefferle]

Every list on a page landed in the page's OPML and was advertised everywhere the site tells readers about its blogrolls. A list that is only of interest on that page, a few tools or references, had no way out of that.

New block setting, "Part of the page for feed readers", on by default. Switched off, the list is left out of the file of the page and out of what the site advertises elsewhere: it is not in ?opml of its page, the page is not listed in the directory or on the front page when all of its lists are like that, and a bare ?opml on such a page loads the page instead of an empty file.

What it keeps is the subscription in its own context: its own ?opml&group=<anchor>, its own rel="blogroll" link on its page, and the download link under it, which now points at its own file.

Opml::extract_groups() keeps every list, so the files of the single lists are unaffected; Opml::shared_groups() is the new basis for the page file, the page's own discovery link and the index. Index::has_blogroll() stays the cheap string search for every request, and the parsing question is Index::has_subscribable_blogroll(), asked on save and when the directory is built.

Seven tests. Checked on a real page: the page file has only the shared list, ?opml&group=tools has the other one, the head links to both lists and to the page, and both download links point at the right file. On a page whose only list keeps to itself: only that list is advertised, ?opml loads the page, and the page is not in the directory.