First crack at serialization with backrefs.

[richardkiss]

No description provided.

[arvidn]

I think a more challenging bomb is where each leaf is a unique (but large) subset of the atom. But I suppose the python implementation might not deduplicate substr.

[richardkiss]

Deduplicating substr is a much bigger challenge, since the number of substrings in a string of length n is $O(n^2)$. We'd have to hash all of them, very tricky. Not sure how zlib does it; maybe it just builds a reverse table of prefixes. It's not in the scope of this change for sure.

[ChiaMineJP]

Small suggetion.
I think adding information showing that left=0 and right=1 is helpful.
Like

"""
Convert a list of 0(left)/1(right) values to a path expected by clvm.
...
"""

[ChiaMineJP]

For consistency, test_object_cache.py should be renamed to object_cache_test.py.

Plus, how about adding read_cache_lookup_test.py?
Following lines to understand how ReadCacheLookup works was a real challenge for me without a testing.
For example, in my understanding root_hash of ReadCacheLookup is a tree-hash representation of read_stack but adding test by original author will greatly help understanding the intention of those code.

[richardkiss]

Renamed.

[richardkiss]

I agree, this is difficult to understand and some tests may help elucidate (and test!).

[richardkiss]

I've added some tests for the ReadCacheLookup class.

[ChiaMineJP]

PyCharm warns that set([a, b, c]) can be replaced with {a, b, c}

[ChiaMineJP]

I think circular dependency of clvm_tools should be avoided.
How about removing assemble and replacing with to_sexp_f here?

class ObjectCacheTest(unittest.TestCase):
    def check(self, obj, expected_hash, expected_length):
        th = ObjectCache(treehash)
        self.assertEqual(th.get(obj).hex(), expected_hash)
        sl = ObjectCache(serialized_length)
        self.assertEqual(sl.get(obj), expected_length)

    def test_various(self):
        self.check(
            # "0x00"
            to_sexp_f(b'\x00'),
            "47dc540c94ceb704a23875c11273e16bb0b8a87aed84de911f2133568115f254",
            1,
        )
        self.check(
            # "0"
            to_sexp_f(None),
            "4bf5122f344554c53bde2ebb8cd2b7e3d1600ad631c385a5d7cce23c7785459a", 1
        )
        self.check(
            # "foo"
            to_sexp_f("foo"),
            "0080b50a51ecd0ccfaaa4d49dba866fe58724f18445d30202bafb03e21eef6cb", 4
        )
        self.check(
            # "(foo . bar)",
            to_sexp_f(("foo", "bar")),
            "c518e45ae6a7b4146017b7a1d81639051b132f1f5572ce3088a3898a9ed1280b",
            9,
        )
        self.check(
            # "(this is a longer test of a deeper tree)"
            to_sexp_f([b"this", b"is", b"\x02", b"longer", b"test", b"of", b"\x02", b"deeper", b"tree"]),
            "0a072d7d860d77d8e290ced0fdb29a271198ca3db54d701c45d831e3aae6422c",
            47,
        )

[richardkiss]

The tests already depend upon clvm_tools as they use brun internally. I agree though, you're right that I should use to_sexp_f.