Hi Alfonso, On Fri, Mar 8, 2013 at 11:06 AM, Alfonso Nishikawa <[email protected]> wrote: >> so, thanks for your answers :) >> I will try to read about beans a bit in the future, my java knowledge >> is just so limited :/ > > Don't worry much. In general we will need later (much later) a full > sourcecode review, I guess. We lack some "Developing a new DataSource" > manual. I want to write some about, but time... > I'm so limited in general... ;P
oh yes, absolutely ;) >>> The real fix is: >>> >>> * Modify HectorUtils.java: >>> - change all "public static<K> void insertSubColumn..." to "public >>> syncronized static<K>..." (3 changes) >>> - change "public static<K> void deleteSubColumn" to " public >>> syncronized static<K> void deleteSubColumn" (1 change) >>> >>> And then you are just synchronizing the calls to the mutator. >> >> Yes, now I think this, too. >> And I think we should really use ->addInsertation (or make it useable >> for CassandraStore) instead of ->insert ... this gives a real >> performance boost. > > Didn't knew about addInsertation(). Good discover. Feel free to open > an issue (and remember to put the "CAVEAT" notice shown over > Mutator#addInsertion()) :) Do you mean the code comment (CAVEAT: a large number of calls with a typo in one of them will leave things in an indeterminant state if we dont validate against LIVE (but cached of course)...) or that these batched mutations are not visible to get requests of even the same mutator (read this somewhere can't remember where)? --Roland
