iClerk refactor
the skip/retain/recalc handlign was confusing. This removes the switcheroo between leveled_codec and leveled_iclerk when mkaing the decision. Also now the building of the accumulator is handled efficiently (not using ++ on the list). Tried to rmeove as much of ?HEAD tag handling from leveled_head - as we want leveled_head to be only concerned with the head manipulation for object tags (?STD, ?RIAK and user-defined).
This commit is contained in:
parent
8e687ee7c8
commit
e0352414f2
4 changed files with 116 additions and 133 deletions
|
@ -1033,10 +1033,8 @@ book_destroy(Pid) ->
|
|||
%% The function will be 1-arity, and can be passed the absolute folder name
|
||||
%% to store the backup.
|
||||
%%
|
||||
%% Backup files are hard-linked. Does not work in head_only mode
|
||||
%%
|
||||
%% TODO: Can extend to head_only mode, and also support another parameter
|
||||
%% which would backup persisted part of ledger (to make restart faster)
|
||||
%% Backup files are hard-linked. Does not work in head_only mode, or if
|
||||
%% index changes are used with a `skip` compaction/reload strategy
|
||||
book_hotbackup(Pid) ->
|
||||
gen_server:call(Pid, hot_backup, infinity).
|
||||
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue