justinli500 commented on code in PR #49855:
URL: https://github.com/apache/arrow/pull/49855#discussion_r3726255968
##########
cpp/src/arrow/io/caching.cc:
##########
@@ -172,46 +181,62 @@ struct ReadRangeCache::Impl {
return new_entries;
}
- // Add the given ranges to the cache, coalescing them where possible
- virtual Status Cache(std::vector<ReadRange> ranges) {
+ // -- Public entry points (acquire entry_mutex, then delegate). --
+
+ // Add the given ranges to the cache, coalescing them where possible.
+ Status Cache(std::vector<ReadRange> ranges) {
ARROW_ASSIGN_OR_RAISE(
ranges, internal::CoalesceReadRanges(std::move(ranges),
options.hole_size_limit,
options.range_size_limit));
- std::vector<RangeCacheEntry> new_entries = MakeCacheEntries(ranges);
- // Add new entries, themselves ordered by offset
- if (entries.size() > 0) {
- std::vector<RangeCacheEntry> merged(entries.size() + new_entries.size());
- std::merge(entries.begin(), entries.end(), new_entries.begin(),
new_entries.end(),
- merged.begin());
- entries = std::move(merged);
- } else {
- entries = std::move(new_entries);
+ Status st;
+ {
+ std::unique_lock<std::mutex> guard(entry_mutex);
+ std::vector<RangeCacheEntry> new_entries = MakeCacheEntries(ranges);
+ // Add new entries, themselves ordered by offset
+ if (entries.size() > 0) {
+ std::deque<RangeCacheEntry> merged(entries.size() +
new_entries.size());
+ std::merge(entries.begin(), entries.end(), new_entries.begin(),
new_entries.end(),
+ merged.begin());
+ entries = std::move(merged);
+ } else {
+ for (auto& entry : new_entries) {
+ entries.push_back(std::move(entry));
+ }
+ }
Review Comment:
Great point! I went with the move-based merge version - the new ranges can
fall between existing ones, so inserting them all at the front could definitely
cause some issues.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
