Trouble with validating merkle proof

[june-cho-88]

I just picked a random transaction to validate which is 94c8536b564c166286d850f9a93835703463e92403d8267e0bb81117242a808c.

My GetMerkleProofResponse gave these following data:

• Merkle Root

0c9b4118bfddce1211217c414cf5bf7be90365042ae7581c4106f11dbc1c8caa

• 11 Merkle Proof Hashes

[0]12d129303667035e96b34d69fd4bc701ed9b01a20c2c2c8978624505b4f23f9c
[1]3ad3d86e85db9a526651613e305d95df459a63b213f0aa73ad93cc3abb31bc7e
[2]961fad10818b0063f01bd60700c673a6ad3ed955abae509adbd6147506a44028
[3]32a5c9fcb1bd6e36523f24e21c9100d5081f90a3f17301ae2f5ca64db071d0d7
[4]20eacb642554683c12f5bb6fcb00858ce7ce234593f4a4a96f2178bd88d3a65d
[5]b508068ef823ef6285c086c24f22ae06da6816cb1e09915c62e8c1398f6c0ba8
[6]d1de2349af80edda1cc12f5641160884042e11adc225e1e636266264ded4598c
[7]94c8536b564c166286d850f9a93835703463e92403d8267e0bb81117242a808c
[8]f34118013afb1179d4a470e849c748e79459bcf8eeea72db7a990284dc3e56ad
[9]e590dd31a2abc736bd6085527c9ec19da0285427e0b3ac0040ef44d3cd0c8786
[10]0452de19385dc9556526e0e14201cf4a124844f790beba7a6eb673187cfcd467

• Flags

0b101010111011011000000010 (0xABB602)

When I'm trying to build a partial merkle tree based on this documentation, it doesn't match with full merkle tree.
For example, first of merkle proof hashes, [0]12d129303667035e96b34d69fd4bc701ed9b01a20c2c2c8978624505b4f23f9c, should be on a left side of depth 2 according to full merkle tree, but the second flag bit(0) says it is on a left of a root, depth 1.

It seems like there is nothing wrong with bchd or transaction itself.
Did I misunderstand the documentation or something?

[cpacia]

Let me see if I understand it 0b101010111011011000000010

• The first bit (1) means the left hash (we'll call this X) at depth 1 needs to be computed
• The second bit (0) means the left hash for X, at depth 2, is the first hash in the provided list of hashes (0x12d12930)

That looks right to me. Am I reading it wrong?

[june-cho-88]

I thought first bit represents the merkle root that needs to be computed later.
I try to use your approach for the next proof hash, but I'm stuck again.

Here is a complete partial merkle tree that gave me a correct merkle root hash.

Depth 02, Left = proofHashes[0]
Depth 03, Left = proofHashes[1]
Depth 04, Left = proofHashes[2]
Depth 05, Left = proofHashes[3]
Depth 07, Left = proofHashes[4]
Depth 09, Left = proofHashes[5]
Depth 10, Left = proofHashes[6]
Depth 10, Right = proofHashes[7]
Depth 08, Right = proofHashes[8]
Depth 06, Right = proofHashes[9]
Depth 01, Right = proofHashes[10]

[june-cho-88]

Here is a partial merkle tree I built from a full merkle tree with full transactions.

At first, I was following a description from bchd.

Binary representing the location of the matching transaction in the full merkle tree, starting with the root (1) at position/level 0, where 1 corresponds to a left branch and 01 is a right branch.

So I tried to find a sequence of 4 same bits in flag bits since there is left-sided pattern that appeared from depth 2 to depth 5, but I couldn't.
What am I missing?