QCon is A practitioner-driven conference focused on facilitating the spread of knowledge and innovation in enterprise software development, or at least that's what their Twitter account says. This is a post about some of the interesting talks I saw at QCon London 2015.
I've been before, in 2013, and had a great time. The talks at QCon are always some of the most varied of any conference I've been to. Rather than focusing on one area there are many tracks of talks with individual talks ranging from management techniques and agile processes to low level bit-packing binary messages for nano-second performance gains.
Instead of going to QCon in 2014 I went to Devoxx instead, which was also a great conference. But Devoxx is entirely Java focused and while I am a full time Java developer I do a enjoy some variety and there was only so many times I could hear about Java 8 Streams at the 2014 conference!
March this year it was back to QCon. I won't list everything here but here's a brief rundown on some of the great talks I went to:
- Treat your Code as a Crime Scene - Adam Tornhill
- This session was all about how you can leverage techniques used in crime forensics to surface data about your code using resources we already have. The end result was a suite of tools and techniques you can use to analyse your version control to find out which areas of code are likely to have bugs or highlight when teams and projects might not be organised in the best way. More information can be found in his book Your Code as a Crime Scene or with his tools
- Modelling complex game economy with a graph database - Yan Cui
- Priming Java for Speed at Market Open - Gil Tene
- Gil Tene is one of my favourite speakers when it comes to Java. He's the CTO and co-founder of Azul Systems, who make a proprietary JVM so he always has some fantastic low level insights for me as a Java developer. This year he was talking about how the JVM optimises your code via JIT compilation, which most people know. Unfortunately most people don't know about how those optimisations can be thrown away if different branches start to be taken. The crux of his talk was how trading companies cannot prime their JVMs with test data before market open to have them ready to run at peak speed because when they flip the switch to live data, previously optimised code paths can come undone. The JVM Azul sell however can now record the statistics used to make the optimisations with live data from one day, and pre-prime the JVM with that information for the next day.
One of my colleagues from Fivium, Ben Basson who also went to QCon this year, also wrote a blog post with a lot more detail and information about the talks he went to which is worth a read too.
The conference wasn't all talks however, there was also a couple of exhibition halls with companies to talk to. Hazelcast had a stand with some great people on it willing to discuss their product and how we might integrate it. Perforce were also there, and as we use Perforce for our version control needs at Fivium I spent some time talking to them to hear what new tools they're coming up with and express my dislike for some bugs we find in their software.
I also put my business card in a box at the Perforce stand and ended up winning a raffle draw for a Sonos PLAY:1 speaker which made the conference a lot more memorable than usual!
After a long 3 days of conference attendance, nothing beats a nice short walk home along the river with some nice London skylines :)