Give it the XML, get back the entries. Fetching is the caller's business: kastrup shells out to curl, the nomad phone app uses its own HTTP client, and neither belongs in here.

for item in feed::parse(&xml, "isene.org", url) {
    println!("{} — {}", item.title, item.link);
}

Not an XML parser

Deliberately. Feeds in the wild are ill-formed often enough that a strict parser is a liability, and the shape being read is shallow: find a tag, take what is between it and its close. CDATA and the five XML entities are handled; namespaces are ignored, which is exactly why dc:creator and content:encoded work by name.

Dates arrive in two spellings — RFC 2822 (Thu, 3 Apr 2026 09:15:00 +0200) and ISO 8601 (2026-04-03T09:15:00Z) — and both are read, with their zone offsets, into Unix seconds. A feed that gives no usable date yields 0 rather than today, so a sort never lies about it.

Identity

An item's id is its own guid/id if it has one, else its link, else its title. Feeds reorder and repaginate constantly, so position means nothing, and an id that changed with the link would orphan the entry every time a site tidied its URLs.

License

Public domain. See Unlicense.