It's a fair point: yes is a professor of the Unimilano university, but is the whole university (similar to MIT) that collaborate to maintain it. He has built the template to autogenerate the different types of collections and the library itself exists from many years (> 10 years), used especially in the low latency space due to the low footprint and predictable performance of collections. In case we believe isn't going to be maintained (very unlikely after so many years) we can return back to use java common collections, given that the signature is the same most of the time.
Il ven 1 mag 2020, 13:12 Christopher Shannon < [email protected]> ha scritto: > This library just seems to be only maintained by a single person, not sure > we really want to tie ourselves to a library like that which would have a > high likelihood of becoming unmaintained at some point. > > On Fri, May 1, 2020 at 6:35 AM Krzysztof <[email protected]> wrote: > > > Yes, that's a shame that Java doesn't have out of the box (no pun > intended) > > support for collections of primitive types. > > > > +1 > > > > On Fri, May 1, 2020 at 12:13 PM nigro_franz <[email protected]> wrote: > > > > > Hi, > > > > > > Recently I am struggling to reduce the memory footprint while loading > the > > > Artemis journal, mostly due to a Java limitations of providing > efficient > > > data-structures using primitive types. > > > In particular I would like a linked hash map implementation using long > > keys > > > and I am aware only of http://fastutil.di.unimi.it/ that implement it. > > > I think we have many parts in the code that would benefit from saving > > > Long/Integer boxed types to lower the memory footprint of the broker, > but > > > such library is a 18 MB jar...!!!! > > > I am +1 to use it and will soon send a PR to show the benefits but I am > > > open > > > to discuss/vote any opinion > > > > > > Thanks, > > > Franz > > > > > > > > > > > > -- > > > Sent from: > > > http://activemq.2283324.n4.nabble.com/ActiveMQ-Dev-f2368404.html > > > > > >
