ruby user group berlin: JRuby, YARV/1.9, and more

March 3, 2007

We had our 2nd ruby users group berlin meeting yesterday with two speakers and the demo of the yet to be released project.

First was Tim Lossen giving a good round-up of the JRuby developments. Not of much interest to me because i have’t touched Java in a while. There was a common understanding that JRuby is a good thing and will pave Rubys way into the enterprise world, and with Sun now as official backing partner, JRuby is heading for a 1.0 release this summer for Javaworld confererence. You can already run JRuby based Rails applications inside you IBM Websphere Application server, Yeah! But can you run a Rails application with JRuby from inside a Java applet on your client browser? Hm, interesting idea, we couldn’t answer that yesterday.

Next was the talk by Murphy about the state of the ruby 1.9 release. Murphy mainly used Mauricio Fernandez eigenclass for reference and gave a really great overview around the three main themes of this topic: Roadmap, New and changed features and performance. Everybody loves the hand drawn roadmap image(which i can’t find now) and while a Ruby2.0 release being something from a far utopian future, we might see a 1.9 release later this year. I’m actually not following the 1.9 developments but became inspired to check again. Enumerators for examples reminded me to my STL/C++ years, just now without the template pains :-)
Interessting were his comments on performance. Tim already showed some charts which related the JRuby to some other implementations and Murphy made some own benchmarks which were pretty much in line with Tims data. The general information is that 1.9(==YARV) is a couple of times faster, ranging from 3 to 10 times faster. BUT! and that is a big but, Murphy did report that on the real life applications he tested, the speed-up was close to insignificant for various applications. This is because the the performance tuning in 1.9 seems to be focussing on benchmark relevant stuff. And real life application are hardly build from benchmark functionallity. This sounds like, been there, seen that before. History(benchmark tweaking) is repeating itself. For me it doesn’t matter. When others can do 4000 requests/second, ruby/rails is definitly fast enough for me.

Finally Benjamin Krause showcased his upcoming OMDB project(tech blog, development version, live). OMDB is a IMDb in wikipedia style with a creative commons licence. 16501 People(see comment) 16000 movies are already in the database and once it will open up, everybody can extend it. Thats a cool idea conceptually and what he showed technically was nothing less than the equivalent to an “Full House” in poker. For example the subsecond async response times for fetching actors from a huge database which were made possible by his ferret magic. impressive.

And this also led to the agenda for the next meeting where Benjamin will give a talk about ferret on Rails. Everybody wanted to see more of it. Also we will have a talk by Adam about AmzonWebServices: S3 and Rails on EC2 . I’m looking forward to it. And about the open mic section, i’m pretty sure we are releasing our foto-foo into the wild.

And for you to have some fun, we plan to record the talks next time and put them up as podcasts to fit with your online consumption habits. Murphy and Tim also promised to upload their talks for online viewing (to the wiki i guess).

looks like the ruby users group berlin is consolidating.

Technorati Tags: , , , , , , , ,