zcash_client_backend: Implement note management via change splitting. by nuttycom · Pull Request #1579 · zcash/librustzcash

nuttycom · 2024-10-17T01:42:39Z

Best reviewed commit-by-commit, hiding whitespace changes.

Part of #1355.

codecov · 2024-10-17T02:01:55Z

Codecov Report

Attention: Patch coverage is 65.62009% with 219 lines in your changes missing coverage. Please review.

Project coverage is 56.40%. Comparing base (791b371) to head (47b1065).
Report is 40 commits behind head on main.

Files with missing lines	Patch %	Lines
zcash_client_backend/src/data_api/wallet.rs	54.45%	46 Missing ⚠️
zcash_client_backend/src/fees/common.rs	64.00%	45 Missing ⚠️
zcash_client_backend/src/data_api/testing/pool.rs	75.00%	42 Missing ⚠️
...ent_backend/src/data_api/wallet/input_selection.rs	39.39%	40 Missing ⚠️
...client_backend/src/data_api/testing/transparent.rs	0.00%	12 Missing ⚠️
zcash_client_sqlite/src/lib.rs	36.84%	12 Missing ⚠️
zcash_client_backend/src/data_api/testing.rs	72.22%	5 Missing ⚠️
zcash_client_backend/src/fees.rs	77.27%	5 Missing ⚠️
zcash_client_backend/src/data_api/error.rs	20.00%	4 Missing ⚠️
zcash_client_backend/src/data_api.rs	66.66%	3 Missing ⚠️
... and 3 more

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1579      +/-   ##
==========================================
+ Coverage   55.93%   56.40%   +0.46%     
==========================================
  Files         149      149              
  Lines       18728    19078     +350     
==========================================
+ Hits        10476    10760     +284     
- Misses       8252     8318      +66

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

nuttycom · 2024-10-17T22:33:16Z

force-pushed to fix the name of quotient in DivRem.

str4d

Flushing comments from review up to 5738990 (I'm still reviewing the last commit).

str4d

Reviewed 6c95fd8.

daira · 2024-10-18T15:24:39Z

+    let mut found_tx_change_memo = false;
+    let mut found_tx_empty_memo = false;
+    T::with_decrypted_pool_memos(&d_tx, |memo| {
+        if Memo::try_from(memo).unwrap() == change_memo {
+            found_tx_change_memo = true
+        }
+        if Memo::try_from(memo).unwrap() == Memo::Empty {
+            found_tx_empty_memo = true
+        }
+    });
+    assert!(found_tx_change_memo);
+    assert!(found_tx_empty_memo);
+
+    // Verify that the stored sent notes match what we're expecting
+    let sent_note_ids = st
+        .wallet()
+        .get_sent_note_ids(&sent_tx_id, T::SHIELDED_PROTOCOL)
+        .unwrap();
+    assert_eq!(sent_note_ids.len(), 3);
+
+    // The sent memo should be the empty memo for the sent output, and the
+    // change output's memo should be as specified.
+    let mut change_memo_count = 0;
+    let mut found_sent_empty_memo = false;
+    for sent_note_id in sent_note_ids {
+        match st
+            .wallet()
+            .get_memo(sent_note_id)
+            .expect("Note id is valid")
+            .as_ref()
+        {
+            Some(m) if m == &change_memo => {
+                change_memo_count += 1;
+            }
+            Some(m) if m == &Memo::Empty => {
+                found_sent_empty_memo = true;
+            }
+            Some(other) => panic!("Unexpected memo value: {:?}", other),
+            None => panic!("Memo should not be stored as NULL"),
+        }
+    }
+    assert_eq!(change_memo_count, 2);
+    assert!(found_sent_empty_memo);


I think nothing here checks that it is specifically the sent note (that is, the one with value 100_0000 zatoshis sent to the to address) that has the empty memo.

nuttycom · 2024-10-20T19:58:59Z

force-pushed to address comments from code review.

Then,

force-pushed to rebase on main.

pacu

LGTM!

left some non-blocking suggestions

pacu · 2024-10-22T19:32:40Z

+            );
+
+            // If we don't have as many change outputs as we expected, recompute the fee.
+            let (fee_with_change, excess_fee) =


[non-blocking] This section of the code from L. 450 to L. 526 is pretty hard to read and follow (even with the generous comments). I wonder if there's a way to reduce its complexity in terms of lamdas and if-else statements so not only is more friendly to follow and maintain, but also to test.

… instead of a tuple.

…allet metadata. In the process this modifies input selection to take the change strategy as an explicit argument, rather than being wrapped as part of the input selector.

…ance`

Co-authored-by: Jack Grigg <thestr4d@gmail.com> Co-authored-by: Daira-Emma Hopwood <daira@jacaranda.org>

nuttycom · 2024-10-23T17:24:46Z

force-pushed to rebase on main to fix changelog conflicts.

str4d

utACK 47b1065

daira · 2024-10-25T04:04:47Z

+    };

    // Once we calculate the balance with and without change, there are five cases:
    //


"change output is added" -> "change output(s) are added" on line 337.

Also on line 345, it's unclear how many change outputs will be added in that case, so the comment should be clarified (there only needs to be one for the purpose of including a change memo, but that doesn't necessarily preserve indistinguishability).

daira

Post-hoc ACK with comments (please address them in #1590).

Co-authored-by: Daira-Emma Hopwood <daira@jacaranda.org>

nuttycom force-pushed the wallet/multi_output_change_strategy branch from 6387785 to e426b68 Compare October 17, 2024 02:12

nuttycom commented Oct 17, 2024

View reviewed changes

Comment thread zcash_client_backend/src/data_api/wallet.rs

nuttycom force-pushed the wallet/multi_output_change_strategy branch 6 times, most recently from 36a54aa to e6c1fa3 Compare October 17, 2024 20:12

daira reviewed Oct 17, 2024

View reviewed changes

Comment thread components/zcash_protocol/src/value.rs Outdated

nuttycom force-pushed the wallet/multi_output_change_strategy branch from e6c1fa3 to 6c95fd8 Compare October 17, 2024 22:32

str4d reviewed Oct 17, 2024

View reviewed changes