Last weekend Bruno and I attended JRubyConf 2012 and its parent event, eurucamp 2012. The conference took place in the outskirts of Berlin, at a beautiful venue on the shore of Lake Müggelsee.

It was a fun weekend, full of interesting sessions, interesting people and interesting activities. The organizers went out of their way to make sure everything went smoothly even in the face of difficult problems such as their WiFi provider not providing enough bandwidth for everyone (they ended up getting a pair of access points to connect us to the Net) and the hotel kitchen apparently misunderstanding what “vegetarian” means. The conference itself was different from others I’ve attended in that it was a bit more laid-back. For example, on Saturday there was a 5-hour lunch/siesta break with activities in and around the lake, such as a coffee-brewing workshop and a martial-arts course that promised to show you how to kill people with your cellphone. This atmosphere made the conference more enjoyable and less stressful than others I’ve been to. Aside from Friday where there were three distinct tracks (JRubyConf, Beginners Workshops and Advanced Workshops), the rest of the conference was a single-track affair, which made it a lot easier to follow everything and prevented scheduling conflicts such as seen in other conferences.

One interesting feature was that all attendees were provided with an RFID tag. These tags served two purposes: on the one hand, there were RFID receivers spread throughout the conference hall which tracked all attendees (as long as they were inside the hall – those outside were shown as grouped around the window). On the other hand, the tags talked between themselves and periodically uploaded the information, which served to generate a graph showing who talked to whom. Many congratulations go to the efforts of Norbert Crombach who built up a couple of visualizations of the data. Sure beats exchanging business cards! There’s more info on the RFID project (which was sponsored by Neo4j) in this blog post

Now to the sessions. Here is the full session schedule. On the first day we attended the JRubyConf track. In the keynote, Tom Enebo gave us the skinny on JRuby’s present and future, which looks brighter every day. There were talks on concurrency and performance issues, benchmarking and even a talk about JRuby on Android (Ruboto). There was so much information that things started running together, so in this post I’ll concentrate on a few highlights from some of the talks that stuck to my mind. The full presentations are currently being uploaded to the JRubyConf and eurucamp Lanyrd sites, and there is a Flickr group where attendees are uploading their pictures.

• David Tollmyr showed us that java.util.concurrent is your friend, and that there are several common operations like Array.join with non-string data or passing binary RubyStrings to Java that can blow up your code in interesting ways or simply make your performance suck. There was a lot more info regarding surprising effects on performance of common Ruby idioms.

• Theo Hultberg talked about HotBunnies, his JRuby RabbitMQ wrapper, and went on to say quite a bit more about java.util.concurrent and how, in a distributed system, blocking is not always a bad idea. He also mentioned Mikka, a thin Ruby wrapper around the Akka library that provides support for actor-based concurrency in JRuby. This meshed in a bit with my talk, where I also talked about Akka and mentioned Mikka.

• Uwe Kubosch talked about Ruboto, the JRuby port for Android. It is currently usable, but startup time and performance still need quite a bit of work. It looks like a great way to write Android apps with a lot less boilerplate than the standard SDK and I will be following it closely.

The last JRubyConf talk was about integrating JRuby and Scala and was delivered by none other than yours truly. I was supposed to go just before lunch but since the morning talks had gone long and we were running behind schedule, Alex asked me if we could move me to the end, carving out a slot from the lightning talks to give enough time for my session. Polyglot programming was a recurring topic in several eurucamp talks, so it came in just at the right time. Bruno was kind enough to record the full session which we’ve posted in Vimeo and the slides are in SlideShare. It wasn’t a good time to try to start a religious war, since most of the attendees’ brains were saturated with info from the rest of the day, but several people approached me later to talk about Scala and polyglot programming. All in all I would call it a success.

Static vs Dynamic Typing – Why Not Both? from abstracc on Vimeo.

As I mentioned, there were several more talks, so have a look at the sites above to see the abstracts and look at the slides. After that we had dinner and drinks for a couple of hours, before the eurucamp keynote. Delivered by Steve Klabnik, it was a thought-provoking talk on epistemology, logic, metaphysics, philosophy and aesthetics, and how the pertain to development. The concepts he presented aren’t something you can really absorb in a 45-minute talk, they have to percolate in your brain for a loooong while. This is one of the talks that I would love to have video of so I could review it at leisure.

The last (but not least-important) activity on Friday was the party. Conversation flowed freely, beer not as freely but in enough quantities to lubricate the conversation along, and so ended a great first day.

The second day kicked off with a bang, with Konstantin Haase talking about how to implement an almost-Sinatra clone in 8 lines, but really was about obfuscating and WTFing Ruby code. One very interesting talk (at least for me) was Lorens Naudé‘s session about ZeroMQ, which, despite its name, is not really a message queue but a messaging layer on top of different kinds of sockets (in-process, IPC, TCP or Multicast). I am really looking forward to implementing it in one of our projects and possibly replacing AMQP, since ZeroMQ seems to be much better suited for our use case.

Then James Coglan went on the stage to talk about Dependency Injection, and how it’s not really for testing but for usability, and that usability leads to testability. This is a very interesting twist, with lots of food for thought.

Some more highlights from day 2:

• Laurent Sanzonetti showed us RubyMotion, a tool to allow you to program iOS devices in Ruby without wrapping your head around Objective-C. Very nice. Since Apple doesn’t allow interpreted code to run on their devices, RubyMotion actually compiles Ruby down to binary before generating the package. It’s a for-pay tool, but if you’re doing any iOS development at all it’s worth a look.

• Srdan Kvrgic talked about how idempotency can help you scale and recover when bad things happen, as long as your idempotency store is reliable and can keep up with the rest of the system. He was the third speaker from Burt and it seems that these guys REALLY know what they’re talking about regarding scalability. We had a couple of chats with them during the lunches and they are working on very cool stuff.

• Piotr Solnica talked about DataMapper 2. This looks like a VERY interesting persistence framework that promises to do away with the kitchen-sink mess that is ActiveRecord. It will even unify SQL and NoSQL backends, with links between them. Still a work-in-progress in progress, but we’ll be watching closely.

Lunch on day 2 was a 5-hour lunch/siesta break with packed lunches. People signed up for activities using hashtags on Twitter (very cool) and could check out how many people had signed up for each activity on a real-time web page the organizers whipped up on the spot. There was a martial-arts course (which I mentioned at the beginning), bike rides around the lake, ultimate frisbee (whatever that is), a coffee-brewing workshop, and (as expected) a Ruboto workshop. Of course, you were free to just rest or, like we did, have a stroll around the lake with into the small village and have a drink there next to the lake. For dinner, some people stayed at the hotel and went bowling, and others went to the village to have dinner at a restaurant there. In the end we had to split into two groups due to scalability limitations of one restaurants’ kitchen. It was a great evening where we wound up talking about functional programming and monads after having a couple of beers, always an interesting prospect.

Day 3 started with a rousing discussion between Isaak Wolkerstorfer and his wife Joan about science, art, beauty, programming and the relationships therein. It was great and really kept the audience’s interest, a hard thing todo especially at 9:30 AM on a Sunday after a late evening the previous night. Another talk which would have been great to have access to the video of.

Then we got onto the more deeply technical discussions. There were a few lightning talks followed by Pablo Astigarraga who came all the way from Uruguay, talking about how there aren’t really any hard and fast rules as to whether software should be monolithic or distributed, and how to prepare for the inevitable architectural shift that comes at some point between “we need it out there quick, monolithic development is faster” and “we have so many users that a single monolithic application cannot keep up”.

Mike Burns came up to talk about programming paradigms other than Object Orientation, with a good bit of Haskell thrown in. His was an extremely interesting talk with lots of food for thought. I am already partly on that road, using functional techniques more and more, but the process of rewiring your brain sure takes time!

Then came another polyglot talk, with Txus Bach, one of the organizers of BaRuCo speaking about polyglot programming in Rubinius. His talk had some things in common with mine, although his was more about implementing his own language whereas I talked about integrating multiple existing ones. Two sides of the same coin.

The conference finally ended with an extremely entertaining talk by Nick Sutterer, who talked about Rails and avoiding the standard mindset associated with it in the interest of testability and cleanliness.

The awesome speakers

All in all it was a great conference. The only thing I really miss are having videos available of all the sessions (especially the more thought-provoking and intriguing ones), but given that it was really organized in a couple of months and that the price was extremely affordable, it’s understandable.  I’d like to thank Alex K, Alex C, Florian and the rest of the team for being wonderful hosts, and we hope to be there next year!

Tanto las charlas de la EuRuCamp como de la JRubyConf tienen muy buena pinta, más todavía porque han aceptado mi charla “Static or dynamic typing? Why not both? Integrating JRuby and Scala” en la cual hablaré sobre la integración entre JRuby y Scala.

Más allá de las guerras religiosas, tanto los lenguajes estáticos como los dinámicos tienen sus ventajas y desventajas. Una de las grandes ventajas de programar sobre la JVM es que hay una gran cantidad de lenguajes con distintas fortalezas y paradigmas de programación. ¿Por qué no aprovechar, y utilizar la herramienta adecuada para cada labor?

Mario Camou – talk from Mario Camou on Vimeo.

Por cierto, el día de hoy, jueves 9 de agosto, y bajo los auspicios del Madrid JUG, daremos una vista previa de la charla en nuestras oficinas. La cita es a las 19:30. Para más información mirad en la página de eventos del grupo.

Como podéis ver, en abstra·cc nos encanta formar parte de la gran familia global de desarrolladores de software. ¡Y presumir de ello!

Hubo una buena concurrencia, aparte de cuatro miembros de nuestro equipo de desarrollo, asistieron varios miembros del JUG Madrid.

En la charla, Miguel Ángel nos presentó tanto la historia y razón de ser de OSGi, como algunos ejemplos prácticos de cómo hacer y utilizar componentes. Además de ser una excelente forma de agregar modularidad a las aplicaciones Java, proporciona características como la actualización en caliente de componentes de la aplicación sin tener que reiniciarla, y el reemplazo, también en caliente, implementaciones de componentes (por ejemplo, para utilizar distintos drivers de Base de Datos), y la construcción de aplicaciones bajo un modelo de Servicios (al estilo SOA pero dentro de la misma aplicación). OSGi fue pensado como una solución para software empotrado, pero sus funcionalidades han crecido con el paso de los años. Existen además varias implementaciones de OSGi, de las que las más populares son Eclipse Equinox y Apache Felix.

Aunque en Java existirá la posibilidad de utilizar Jigsaw para modularizar nuestros proyectos, el alcance de éste es bastante menor que el de OSGi (está pensado para modularizar las librerías internas del JDK), además de que Oracle ha anunciado el retraso de Jigsaw hasta Java 9.

Actualmente existen bastantes proyectos que utilizan OSGi para su extensibilidad, como por ejemplo Eclipse y Glassfish, así como el gestor documental Nuxeo.

Agradecemos a Miguel Ángel el habernos dado la charla, cuyos slides se pueden encontrar en su GitHub, así como en SlideShare.

Unlike the web admin interface (unsurprisingly known as Gyokuro·admin), our new product will allow end-users to know the status of the whole system:

• Bike and station usage and statuses, including charge levels (if electric bikes are used, that is)
• Important notices regarding service outages
• News or just stuff the service provider considers important
• Information for the general public, regarding fees, legislation and advices for a safe ride

Moreover, if the system is opened for public registration, users will be able to recharge minutes via PayPal or credit card, depending on the business model the customer chooses.

But not so quick, we’ve got just a working, keynote-ready lorem ipsum app that will be ready for your mischief in a couple of weeks. For now, have a peek. Click and see! (oops! only in Spanish for the moment)

We hope you like it as much as we do. We are really enjoying having a break outside our usual enterprisey-, ERP-related stuff. Gyokuro FTW!

Más info sobre el mapa y demás en este post de nuestro blog de desarrollo.

Notas sobre los parámetros de configuración:

Negrita - parámetros prácticamente obligatorios, o por los que se puede comenzar
Normal – parámetros para ajustes finos
Rojo - parámetros que es mejor no tocar (o tocar sólo si sabes muy bien lo que estás haciendo)

• Have you programmed in C++?
• Do you have an interest in Qt?
• Do you like us?
• Would you like to be a part of an exciting product release?

We offer you an opportunity to join our team and help us develop the BlueMountain platform launcher. Besides this summer project, the smart people in our team would love to mentor you in other areas, such as Ruby, functional programming, devops stuff and agile methodologies.

As for the gory details, you’d earn €600/month, working 6 hours a day. A valid cooperation contract has to be signed by us and your University.

Apply now! Just use our contact form or ping us at email hidden; JavaScript is required

This won’t be your typical internship ;-)

A finales del año pasado, se anunció el fin del uso gratuito de la API de Google Maps por parte de los desarrolladores. Sus empresas tendrían que pagar por el acceso a este servicio. Las reacciones más sonadas a este cambio de política de uso, fueron que Apple y Foursquare se pasaron a una alternativa llamada “Open Street Map”, hasta la fecha poco conocida entre los usuarios de apps. Poco después, Wikipedia se suma al cambio, tras anunciar que sus aplicaciones móviles abandonarían Google Maps.

OpenStreetMap es una iniciativa cuyo objetivo es crear y proporcionar datos geográficos libres, tales como callejeros y mapas de carreteras, a todo el mundo. En Open Street Map, los datos de los mapas son libres para su descarga y utilizan una licencia de uso abierta, lo que permite que usuarios y desarrolladores puedan utilizar los mapas de forma más flexible y libre, tanto para crear mapas personales como para añadir información.

Aunque, por ahora, Google Maps sea un producto mucho más desarrollado y útil para los usuarios finales, las empresas que, como abstra·cc, utilizamos Open Stret Map en nuestras aplicaciones, debemos contribuir al proyecto. No sólo usarla porque es gratis. Hay mucho más en juego que el dinero o maniobras comerciales.

En poco más de seis meses, el número de usuarios registrados en Open Street Map, ha superado los 600.000. A día de hoy, Wikipedia tiene 16.643.482.

• http://www.openstreetmap.org/
• http://www.openstreetmap.es/
• @openstreetmapes

• OpenJDK 7 será el entorno Java preferido por Ubuntu Server, a partir de la versión 12.04 LTS (Precise Pangolin), también anunciada ayer.
• Oracle ya distribuye JDK7 para Mac OS X (sólo Lion y posteriores), ofreciendo una alternativa a los paquetes de openjdk-osx-build. Merece la pena recordar que Apple dejó entrever hace un par de años que dejaría de desarrollar su propia plataforma Java, y que cederá a la comunidad las mejoras y características desarrolladas por ellos.

Para nosotros son buenas noticias, ya que podremos garantizar que nuestros productos de gestión funcionarán mejor y más rápidamente en más escenarios.

Además, Canonical también apoya y proporciona soporte para las derivaciones de Ubuntu: Kubuntu, Xubuntu, Edubuntu y Lubuntu. Así que podemos decir que realmente gozamos de alternativas con la misma calidad, cada una de ellas orientada a un tipo de usuario.

El 26 de Abril de 2012, se pone a nuestra disposición “Precise Pangolin”, nombre en clave de la versión 12.04 LTS. Echa un vistazo:
 

En abstra·cc usamos Ubuntu LTS en nuestros servidores, y es la plataforma de escritorio que siempre recomendamos a nuestros usuarios y clientes.

Descarga (Ubuntu, Xubuntu, Kubuntu, Edubuntu,
Lubuntu)