From 2c9d42ed955705a0207b047918f58ef258dcfa4f Mon Sep 17 00:00:00 2001 From: tv Date: Thu, 19 Nov 2026 04:02:52 +0000 Subject: federation: a corpus per revision, judged alike by every later continuum Nothing checked that judging under a revision stays what it was: a change to the fold that alters what an earlier revision takes, refuses or reads parts nodes running it from nodes that don't, and the tests regenerate their expectations with the code. For each revision this continuum knows, a file in `src/federation/corpus/` holds histories generated at that revision, of every kind the properties generate, each event's bytes with what became of it, by kind and never by a refusal's words, and what the store reads after; its first line names the revision by its ID. It is written once, as the revision is made (`write_the_corpus`, an ignored test run by hand), and never edited after; every later continuum judges it alike (`every_revision_judges_as_its_corpus_says`), which fails where a revision has no corpus yet, saying how to write it. The test plan describes it. --- src/federation/mod.rs | 2 ++ 1 file changed, 2 insertions(+) (limited to 'src/federation/mod.rs') diff --git a/src/federation/mod.rs b/src/federation/mod.rs index 2edc7c4..9444712 100644 --- a/src/federation/mod.rs +++ b/src/federation/mod.rs @@ -11,6 +11,8 @@ pub(crate) mod agent; #[cfg(test)] mod attacks; pub(crate) mod client; +#[cfg(test)] +mod corpus; pub(crate) mod event; pub(crate) mod fold; pub(crate) mod fork; -- cgit v1.3.1