10 April 2018
- Code looks cleaner without them.
The most important thing is not whether you use semicolons or not, but that the whole team uses the client’s standard. If the client doesn’t care, then use them, but make sure you use a linter such as ESLint, or configure and share your IDE’s grammar settings. If all the member of your team are using a JetBrains IDE, e.g. IntelliJ IDEA, then you can automatically share many aspects of the IDE and language configuration.
Among the many distressing statistics that come from Stack Overflow’s review of 2017 – other than that PERL is still alive – was the fact that US developers who use tabs (kind, generous, fighting for all the good in the world types) have a median salary that’s roughly 30% less than the despicable kitten-punching space-preferring developers. Where’s the justice? Still, it’s worth a read…
In the pantheon of great phrases to drop into a sales conversation when the topic turns to software architecture, comes: ‘So, rather than the Lambda architecture, you want a Kappa architecture – where the emphasis is on streamed data rather than batches so you can ditch the data store.’ This is a trend that we’ll see more of – you’re not cutting off options and you can still do batch, if necessary.
Lambda architectures often revolve around Hadoop-based solutions whereas Kappa architectures are often based on streaming (well, micro-batching really) via Apache Spark and using a log concept as the data store rather than a traditional SQL server. Our internal researchers are constantly investigating these architectures (among several others).
Spark is often used is because of its flexibility; it’ll use Hadoop’s YARN if you’ve got it and its plugins support SQL, Streaming, GraphX Graph DB and Machine Learning – that’s a lot of flexibility, especially when you consider it already has great language/API support from Scala, Java, Python and ‘R’. Another benefit is it’s fully supported by Microsoft through its Azure HDInsight service. It started out as a proof of validity of the Apache Mesos file system, but became so impressive in its own right, it totally eclipsed Mesos. There are quite a few Pluralsight courses on Spark to get you started.
BLUEPRINT FOR SUCCESS