Hi, > Yeah, at this point, we need something like MVCC implemented in JDBM 3.0
I am afraid I dont have plans for MVCC in JDBM. Having single transaction greatly reduces complexity and I dont really need it in my desktop app. JK On Thu, Dec 23, 2010 at 11:27 PM, Emmanuel Lecharny <[email protected]> wrote: > On 12/23/10 11:30 PM, Jan Kotek wrote: >> >> Hi, >> >> I am proud to announce stable release of JDBM2. It is 1.0 fork which >> adds lot of stuff, most notably: >> * java.util.Map view to indexes >> * soft cache >> * secondary map (aka foreign indexes in SQL) >> * better speed >> * defragmentation >> * more space efficient >> * under Apache 2 license. > > Great !!! >> >> However it does not solve biggest problem you had with 1.0. There is >> still 'big lock' and concurrency access does not scale well. Readonly >> access is very fast, thanks to soft cache. > > Yeah, at this point, we need something like MVCC implemented in JDBM 3.0 to > get it more scalable. Cache can help, but has its limit. > > An interesting next step :) > >> API and file store format are similar, but incompatible with 1.0 > > Will give it a try and see if it still works well with ADS. > > > -- > Regards, > Cordialement, > Emmanuel Lécharny > www.iktek.com > >
