efa08d4081
the relatively small batch size of 128 was chosen so that if we have
a set of keys like
a/1
a/2
...
a/100000
b
c
list operations would not have to walk 100k keys inside of a/ before
skipping to b. unfortunately, iteration is also used by the metainfo
loop. in that case, it's doing a recursive listing, and so there's
no need to skip large prefixes. thus, we can use a bigger batch
size when recursive listing is requested.
Change-Id: I87cf1ba385b6eb2928c5b7cc5e0f7a8c7bd126d9
|
||
|---|---|---|
| .. | ||
| boltdb | ||
| cockroachkv | ||
| filestore | ||
| postgreskv | ||
| redis | ||
| storelogger | ||
| teststore | ||
| testsuite | ||
| blob.go | ||
| common.go | ||
| iterator.go | ||
| listkeys.go | ||
| listv2.go | ||
| util.go | ||