get_range with heed does not work.

[Boog900]

Bug

heed's get_range function returns incorrect results when a different ordering is used with the DB/table.

Steps to reproduce

Add this:

        assert_eq!(table.keys().unwrap().count(), 257);
        assert_eq!(table.values().unwrap().count(), 257);
        assert_eq!(table.get_range(0..257).unwrap().count(), 257);

After this:

cuprate/storage/database/src/backend/tests.rs

Lines 377 to 386 in f3c1a5c

	let tx_ro = env_inner.tx_ro().unwrap();
	let table = env_inner.open_db_ro::<TestTable>(&tx_ro).unwrap();
	let iter = table.iter().unwrap();
	let keys = table.keys().unwrap();
	for ((i, iter), key) in RANGE.zip(iter).zip(keys) {
	let (iter, _) = iter.unwrap();
	let key = key.unwrap();
	assert_eq!(i, iter);
	assert_eq!(iter, key);
	}

---

I think this is because heed's range cursor compares raw bytes here: https://docs.rs/heed/latest/src/heed/iterator/range.rs.html#416

[gzalz]

@Boog900 I think things might be working and that the test may need a slight refactor. Let me know your thoughts.

#350

[Boog900]

Yeah slightly embarrassing but I went too aggressive on trying to simplify the test case, updated the issue with the new test case that actually fails due to this issue.

[Boog900]

meilisearch/heed#289