DragonX/ObsidianDragon
1
1
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 10 Wiki Activity

perf(history): don't rebuild the O(N) tx display cache on every block for a confirmation-only change

Browse Source 
Performance audit: tx_revision bumps on every new block because each tx's confirmation count rises, so
the History tab's groupByTxid + merge + sort + deep-copy display cache fully rebuilt once per ~36s block
on the UI thread, even when nothing but the counts changed.

Add WalletState::tx_structure_revision, bumped alongside tx_revision EXCEPT when setTransactionsIfChanged
sees a pure confirmation-count delta (new sameTransactionLayout: every row equal ignoring the confirmation
COUNT, but the pending/confirmations==0 state — which drives the pending-first sort and shield merge — must
still match). Every other writer bumps both (conservative default). The History tab's groups cache and
display cache now key on tx_structure_revision; on a confirmation-only delta it patches dtx.confirmations
in place (rows, order and indices into state.transactions are unchanged) instead of rebuilding, and re-stamps
the filter snapshot so "export current filter" stays valid. Displayed confirmations still update every block.

No fund path touched; display/caching only. Test: testTxStructureRevision.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
DanS
2026-10-19 14:44:13 -05:00
co-authored by Claude Opus 4.8
parent 6d58802127
commit df98f98669
3 changed files with 112 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/ui/windows/transactions_tab.cpp
+26 -6
View File
@@ -219,10 +219,11 @@ void RenderTransactionsTab(App* app)
// txid grouping shared by the summary pass and the display pass. Indices only, rebuilt once per tx_revision
// (a sort-combo change re-sorts without regrouping).
static history::TxidGroups s_groups;
static std::uint64_t s_groups_key = 0; // 0 never matches: tx_revision starts at 1 and only grows
if (s_groups_key != contentKey) {
static std::uint64_t s_groups_key = 0; // 0 never matches: tx_structure_revision starts at 1 and only grows
const std::uint64_t structKey = state.tx_structure_revision; // ignores pure confirmation-count bumps
if (s_groups_key != structKey) {
s_groups = history::groupByTxid(state.transactions);
s_groups_key = contentKey;
s_groups_key = structKey;
}
// Summary-card totals: Received / Sent / Mined (counts + amounts). These are three static
@@ -415,13 +416,14 @@ void RenderTransactionsTab(App* app)
// case while the user reads/scrolls) the cache is reused. The result is already sorted newest-first by
// the refresh service, but we re-sort here to apply the "pending first" ordering.
static std::vector<DisplayTx> s_display_cache;
static std::uint64_t s_display_cache_key = 0; // tx_revision the cache was built from
static std::uint64_t s_display_cache_key = 0; // tx_structure_revision the cache was built from
static std::uint64_t s_display_conf_rev = 0; // tx_revision the cached confirmations were last synced to
static int s_display_cache_sort = 0; // sort mode the cache was built with
static bool s_display_cache_valid = false;
static std::uint64_t s_display_gen = 0; // bumped on every rebuild; keys the filtered-index cache below
// Two separate fields, not a mixed hash: small revision/sort values would collide under xor-mixing.
if (!s_display_cache_valid || contentKey != s_display_cache_key || s_sort_mode != s_display_cache_sort) {
if (!s_display_cache_valid || structKey != s_display_cache_key || s_sort_mode != s_display_cache_sort) {
++s_display_gen;
s_display_cache.clear();
std::vector<DisplayTx>& display_txns = s_display_cache;
@@ -524,9 +526,27 @@ void RenderTransactionsTab(App* app)
}
});
s_display_cache_key = contentKey;
s_display_cache_key = structKey;
s_display_conf_rev = contentKey; // freshly built from the current list
s_display_cache_sort = s_sort_mode;
s_display_cache_valid = true;
} else if (s_display_conf_rev != contentKey) {
// Confirmation-count-only delta (setTransactionsIfChanged kept tx_structure_revision): the merged rows, their
// order and the indices into state.transactions are unchanged, so refresh just the count (same rule as the
// build above: a shield shows the larger of its two legs). Sort order cannot change: mode 0 only reads
// conf==0, which the layout check holds constant, and the other modes ignore confirmations.
for (DisplayTx& dtx : s_display_cache) {
if (dtx.is_shield) {
dtx.confirmations = std::max(state.transactions[dtx.send_idx].confirmations,
state.transactions[dtx.recv_idx].confirmations);
} else {
dtx.confirmations = state.transactions[dtx.orig_idx].confirmations;
}
}
// The filtered-index cache keys on s_display_gen (unchanged, correct: filters ignore confirmations) and
// the snapshot's indices remain valid; only re-stamp it so 'export current filter' stays usable.
if (s_filter_snapshot.valid) s_filter_snapshot.txRevision = contentKey;
s_display_conf_rev = contentKey;
}
// The merged + sorted list (rebuilt above only on change). Pagination below runs per-frame (it only