datastore: DomainGetAsOfQuery #2680
Conversation
battlmonstr
force-pushed
the
pr/history_get
branch
from
3a0be9e to
6606785
Compare
battlmonstr
force-pushed
the
pr/history_get
branch
2 times, most recently
from
99f9c30 to
0dfa5fc
Compare
battlmonstr
force-pushed
the
pr/history_get
branch
from
0dfa5fc to
b738738
Compare
| return value; | ||
| } | ||
|
|
||
| std::optional<std::pair<size_t, uint64_t>> EliasFanoList32::seek([[maybe_unused]] uint64_t v, bool reverse) const { |
There was a problem hiding this comment.
[[maybe_unused]] is probably a leftover here
There was a problem hiding this comment.
yes, forgot to remove
| kvdb::EncoderConcept TKeyEncoder1, snapshots::EncoderConcept TKeyEncoder2, | ||
| kvdb::DecoderConcept TValueDecoder1, snapshots::DecoderConcept TValueDecoder2, | ||
| const snapshots::SegmentAndAccessorIndexNames* segment_names> | ||
| struct HistoryGetQuery { |
There was a problem hiding this comment.
HistorySeekQuery would be a better name for both capturing the semantics and keeping consistency wrt Erigon.
There was a problem hiding this comment.
I understand the desire to align with Erigon. I've called it Get on purpose, because I think Erigon's naming of this API is inconsistent wrt mdbx seek operation and should be renamed to Get. In my mind the difference is that Get (if it succeeds) returns an exact value matching the query parameters (so the query is like: SELECT * WHERE A = 123 LIMIT 1), but Seek can return something that is not an exact match (so the query is like: SELECT * WHERE A >= 123 LIMIT 1). We also call it lower_bound in some places. Additionally, in my mind Seek should reposition a cursor/iterator to be able to continue iteration from that point on (like fseek). I consider HistoryGetQuery as a close friend of DomainPut/Get.
No description provided.