add WriteBatch #26
add WriteBatch #26
Conversation
|
Hey @akiozihao. rose already came up with some ideas around this feature in #23 . dont u prefer taking this path? it is also similar to tidwal's WAl, except for creating a seperate struct for each batch. |
|
@amityahav Thank you for your reminder; I have noticed this issue and this problem. I am currently attempting to split |
wal.go
Outdated
| wal.mu.Lock() | ||
| defer wal.mu.Unlock() | ||
|
|
||
| if wal.pendingSize+wal.calSizeUpperBound(int64(len(data))) > wal.options.SegmentSize { |
There was a problem hiding this comment.
wal.calSizeUpperBound can be put in a variable and reused in 311 and 314
segment.go
Outdated
| } | ||
|
|
||
| // WriteAll write batch data to the segment file. | ||
| func (seg *segment) WriteAll(data [][]byte) ([]*ChunkPosition, error) { |
There was a problem hiding this comment.
can u maybe privatize all segment functions that are invoked from within wal's methods? such as WriteAll, Write, Read.. as they arent supposed to be used independently. #22
segment.go
Outdated
| // You can call Next to get the next chunk data, | ||
| // and io.EOF will be returned when there is no data. | ||
| func (seg *segment) NewReader() *segmentReader { |
There was a problem hiding this comment.
Please split this into a single PR
For better review.
This approach may have lower efficiency for short values and can potentially disrupt the segment when failures occur, requiring more testing.