On Tue, Nov 27, 2012 at 11:41 AM, Alex Rønne Petersen <xtzgzo...@gmail.com> wrote: >> In short, the proper way to replace an object is to first fetch the >> existing object, then update the value of what you fetched, then use >> riakc_pb_socket:put to write the new value. The reason for doing this >> is to comply with Riak's consistency system (the vector clocks) - it >> needs to know what value you were starting from in order to overwrite >> it correctly. > > OK, makes sense. Just out of curiosity, though, is updating values via > put without an update a supported scenario in some way or is it > generally considered a Bad Idea (tm)?
This depends on each bucket's configuration of allow_mult. In the default setup, with allow_mult=false, writing without fetching first causes Riak to simply store the new value instead of the old value, with conflict resolution (during read-repair/etc.) done simply by comparing modification timestamps. With allow_mult=true, writing without fetching first causes Riak to store sibling objects. > Does Riak figure out based on the name of the index what type it is? > That is, *_int = integer, *_bin = string? > > Is there any physical limit to how many secondary indexes can be > created for an object? Or perhaps a recommended limit? Answers to both of these are on the poorly-named "Configuring Secondary Indexes" page: http://docs.basho.com/riak/latest/cookbooks/Secondary-Indexes---Configuration/ > When if_none_match is enabled, will riakc_pb_socket:put/3 return ok > regardless of whether the value was stored? No, it looks like it returns an error in that case: 6> O = riakc_obj:new(<<"one">>, <<"two">>, <<"three">>). {riakc_obj,<<"one">>,<<"two">>,undefined,[],undefined, <<"three">>} 7> riakc_pb_socket:put(S, O, [if_none_match]). ok 8> riakc_pb_socket:put(S, O, [if_none_match]). {error,<<"match_found">>} -Bryan _______________________________________________ riak-users mailing list riak-users@lists.basho.com http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
