Digests » 142
Digging thru logs to debug issues? Such a waste of time. Rollbar detects when code breaks in real-time and provides the full stack trace and diagnostic data to help you defeat impactful errors. Debug with data and instantly know what's broken and why. Try Rollbar for free!
this week's favorite
Recently, during a pair programming session, I had a really interesting discussion. We talked about validating incoming payload which was basically a client’s input in a JSON format. We were wondering if there’s one proper place to validate requests in web applications. Multiple ideas appeared so I’d like to present them here.
PostgreSQL has great support for objects stored as JSON. This is useful for those moments when you need to store data that could be variably structured, such as responses from other services’ APIs, or data that frequently travels together within your relational tables.
In January 2017, we presented results to the European Commission on a FP7 project called SyncFree, where we were tasked with a large-scale evaluation of the Lasp programming system. In this evaluation, we spent about 9,000+ EUR running a large-scale experiment on Amazon demonstrating that our runtime system and programming environment could scale to a 1,024 node, fully replicated, CRDT data store, using an application scenario from Rovio Entertainment. We’ve written up the results in a paper at PPDP ‘17, that is available online
Best Elixir frameworks to create applications for web, APIs, bots, command-line scripts, dashboards, etc. The list includes Phoenix, Nerves, Sugar, Hedwig and many more.
A lot of the benefits I see being pointed out regarding BEAM are, despite being things that are really impressive for the time in which they were implemented, no longer distinguishing features of Elixir/Erlang. You can have rolling redeployments (hot reloads) using Kubernetes and a container infrastructure, gRPC can bridge many of the gaps on languages that do not have the same distributed actors capabilities, plus when using containers a lot of the monitoring instrumentation comes for free, it's also still a very battle tested model given that a lot of Google's own infrastructure runs on a similar configuration, and using containers also opens up the possibility of running specialized parts of the architecture that are implemented in other languages running on the same cluster.