Why two size_of_entry tests for gix_index?

[EliahKagan]

This is the gix-index size_of_entry test:

gitoxide/gix-index/src/lib.rs

Lines 241 to 248 in 9ab86a2

	#[test]
	fn size_of_entry() {
	assert_eq!(std::mem::size_of::<crate::Entry>(), 80);
	// the reason we have our own time is half the size.
	assert_eq!(std::mem::size_of::<crate::entry::stat::Time>(), 8);
	assert_eq!(std::mem::size_of::<filetime::FileTime>(), 16);
	}

While is the gix-index-tests::integrate index::size_of_entry test:

gitoxide/gix-index/tests/index/mod.rs

Lines 26 to 33 in 9ab86a2

	#[test]
	fn size_of_entry() {
	assert_eq!(std::mem::size_of::<gix_index::Entry>(), 80);
	// the reason we have our own time is half the size.
	assert_eq!(std::mem::size_of::<gix_index::entry::stat::Time>(), 8);
	assert_eq!(std::mem::size_of::<filetime::FileTime>(), 16);
	}

Is there a need for this pair of separate but identical tests? (They are effectively identical because crate in the first test refers to gix_index, which appears in its place in the second test.)

---

Edit: There is a similar situation in gix-pack, where the pair of tests that test Item<()> and Item<EntryWithDefault> are repeated in two places.

These are the gix-pack cache::delta::tests::size_of_pack_tree_item and gix-pack cache::delta::tests::size_of_pack_verify_data_structure tests:

gitoxide/gix-pack/src/cache/delta/mod.rs

Lines 29 to 33 in 9ab86a2

	#[test]
	fn size_of_pack_tree_item() {
	use super::Item;
	assert_eq!(std::mem::size_of::<[Item<()>; 7_500_000]>(), 300_000_000);
	}

gitoxide/gix-pack/src/cache/delta/mod.rs

Lines 35 to 49 in 9ab86a2

	#[test]
	fn size_of_pack_verify_data_structure() {
	use super::Item;
	pub struct EntryWithDefault {
	_index_entry: crate::index::Entry,
	_kind: gix_object::Kind,
	_object_size: u64,
	_decompressed_size: u64,
	_compressed_size: u64,
	_header_size: u16,
	_level: u16,
	}
	assert_eq!(std::mem::size_of::<[Item<EntryWithDefault>; 7_500_000]>(), 840_000_000);
	}

While these are the gix-pack cache::delta::tree::tests::size_of_pack_tree_item and gix-pack cache::delta::tree::tests::size_of_pack_verify_data_structure tests:

gitoxide/gix-pack/src/cache/delta/tree.rs

Lines 229 to 233 in 9ab86a2

	#[test]
	fn size_of_pack_tree_item() {
	use super::Item;
	assert_eq!(std::mem::size_of::<[Item<()>; 7_500_000]>(), 300_000_000);
	}

gitoxide/gix-pack/src/cache/delta/tree.rs

Lines 235 to 249 in 9ab86a2

	#[test]
	fn size_of_pack_verify_data_structure() {
	use super::Item;
	pub struct EntryWithDefault {
	_index_entry: crate::index::Entry,
	_kind: gix_object::Kind,
	_object_size: u64,
	_decompressed_size: u64,
	_compressed_size: u64,
	_header_size: u16,
	_level: u16,
	}
	assert_eq!(std::mem::size_of::<[Item<EntryWithDefault>; 7_500_000]>(), 840_000_000);
	}

[Byron]

Thanks for bringing this up!

And even though I could have thought that maybe I had the same idea, twice, in the case of gix-index, it's hard to imagine that happened with the much more complex duplicated tests in gix-pack.

My answer is that this must have been some sort of accident causing the duplication, I wouldn't see a reason them to exist twice.

[EliahKagan]

For the gix-pack tests, looking again, it looks like c85a981 (#1237) created the tree module, moved code into it, and this one associated test ended up in both places. I don't know what the cause was in gix-index, though.

For both gix-index and gix-pack, I'll deduplicate them as part of a soon-to-be PR.

[EliahKagan]

I've included the removal of one of each of the duplicated tests in #1687.