| commit | ae6df40822b668e63699e01c166de6739284cda3 | |
| author | Olly Betts <olly@survex.com> | |
| Wed, 21 Feb 2018 04:31:37 +0000 (21 17:31 +1300) | ||
| committer | Olly Betts <olly@survex.com> | |
| Wed, 21 Feb 2018 05:28:12 +0000 (21 18:28 +1300) | ||
| tree | f27effedc092c5a174acbddc3a79c3ffeedc07c0 | treesnapshot (tar.gz zip) |
| parent | 26e378e120669c2d9d32303360dd053a2d38e230 | commitdiff |
[honey] Reduce find_entry() and find_entry_lt() use
find_entry(string()) is the same as find_entry_ge(string()).
The previous pattern of using find_entry_lt() to put the cursor in a
state where next() advances to a particular key is a bit inefficient as
we have to read items before those we actually want; it's also a wasted
cursor seek if the first action is skip_to() instead of next(). So
instead put the cursor after_end() initially, and check for that in
next() and use find_entry_ge() for the particular key at that point.
find_entry(string()) is the same as find_entry_ge(string()).
The previous pattern of using find_entry_lt() to put the cursor in a
state where next() advances to a particular key is a bit inefficient as
we have to read items before those we actually want; it's also a wasted
cursor seek if the first action is skip_to() instead of next(). So
instead put the cursor after_end() initially, and check for that in
next() and use find_entry_ge() for the particular key at that point.