Hi there, it’s been a while!
After some vacationing, the ABEND dump is back. The ABEND dump is the issue where I share the most interesting content I’ve been reading, listening to, and watching lately.
Want to check the previous issue? Read it here: ABEND dump #4
Here it is:
Unit or integration tests?
TLDR:
It doesn't matter!
As long as tests run fast, are easy to be written, and give good feedback, we are fine.
Mauricio Aniche is a reference when it comes to software testing. Go check his book and newsletter and thank me later ;-)
Writing a Faster Jsonnet Compiler
Compilers are the kind of software that amazes me. It’s beautiful to see a human-readable text format being translated into a machine-executable binary. Lately, my work has been writing, reading, and maintaining a lot of jsonnet. While performance so far has not been a problem for me, it was an issue for some clever developers working at Databricks, to the point of rewriting the compiler in Scala aiming to improve the compilation time for their usage, giving birth to sjsonnet.
jsonnet: habemus std.flattenDeepArray
If you have custom jsonnet code to flatten a deep array, soon the standard library will have the std.flattenDeepArray function to do it.
In my last post, I wrote for new programmers still in doubt about choosing a programming language:
And in the last issue of GitHub’s Branching Out newsletter, Damien Katz (creator of Apache CouchDB) gave an answer when asked about a piece of advice he would give to new developers, and his answer resonates with what I think (and is related to the post above):
Don’t be dogmatic and get too married to any style, paradigm, or methodology. They all work, and they all don't work. And success really has a lot to do with the team and the team cohesiveness and how well they've bought into whatever it is. The culture and psychological safety in the team—those are the kinds of things that actually make projects succeed or fail, more so than any methodology or technology.