Thursday there was the monthly Scala Central meetup (the number 8). It was hosted by The Telegraph, so we had also a nice view of the journalists working late in the evening. Moreover their offices are close to the Victoria station, so it was really in the center of London.
After a very small introduction about The Telegraph and their technology stack (there was also a Google Home as a prize among the ones who followed the twitter account of their engineering division, but I thought it was not assigned immediately, so I didn’t do it) the first talk was by Giuseppe Scopelliti (LinkedIn – Twitter) who talked about the expression problem, for which the challenge is to define a data type with cases and add new cases of the type, and new operations (of course in Scala). This may be useful if you want to wrap a third party library and add functionalities to all its child-classes.
The trick was to use a feature of Scala called Shadowing that replace one class/variable/trait definition with another one. Basically it means that you can put in scope the same element with the same name and the one closest o the scope is the only visible. With that, traits and abstract classes, you can wrap a class with a class with the same name but extendable.
In the second talk, Gerard Maas,(Twitter – Github – LinkedIn) Lead developer at Kensu, spoke about Scala, Spark and the Spark Notebook. Spark Notebook is an application that let you execute commands in Scala or Spark by a webapp and it gives you back the result. He made a demo of that, and that he faced a problem of classification of different languages (English, Italian, French…) based on long books of the European Parliament (standard books, basically translated in all languages) to show analogies among different languages. To do so he considered ngrams. NGrams are sequences of letters, not single words. He found some analogies, for example the number of usages of the article the and te equivalent in another language.
The third and last talk was by Antonio Rodrigues (LinkedIn – GitHub) and Daniel Conde Garcia (LinkedIn): they spoke about what they did to improve the pipeline to deploy in prod. Basically they started from a pipeline that needed some human interaction and some synchronisation with the devops for the infrastructure to end up with a fully automatic delivery pipeline based on SBT and AWS. The demo failed several times for stupid reasons, but at the end, in the working try, for a basic project created from scratch on github, they took 5 minutes to deploy it in prod.
Before and after the talks, as usual, beers, food and lot of chances to connect to other people.
This time I took a couple of days to write down this post (with the current weather it is difficult to write 😛 ), so the videos are already available. You can find them here.