miniblog.

Emacs tip of the day: annotate your functions! The byte compiler can replace entire function calls at compile time if your function is pure.

Macros where you have to specify a var name are really clean. They have few hygiene issues, and work with both lexical and dynamic binding!

Interesting, well executed research on test-first: https://neverworkintheory.org/2016/10/05/test-driven-development.html (found no benefit, shame to see how defensive people can be)

What else are we getting wrong? https://danghica.blogspot.com/2016/09/what-else-are-we-getting-wrong.html (on the importance of evidence-based PL research)

Amazing Emacs package of the day: https://github.com/emacs-pe/docker-tramp.el allows you to access Docker containers over Tramp!

Good docs cover a multitude of API design sins.

Buttery Smooth Emacs https://www.facebook.com/notes/daniel-colascione/buttery-smooth-emacs/10155313440066102/ (a blunt but fair summary of Emacs' heritage)

Excitingly, the talented coreboot folks have found a way of disabling much of Intel's Management Engine:

How much of a typical GNU/Linux system is GNU software these days? AFAICS gcc and coreutils are the only bits that are ubiquitous.

Amazingly, Linux kernel devs are discussing software mitigations for rowhammer exploits!

I'm increasingly persuaded that deploying web services is simpler with Docker. Debugging Ansible playbooks is slow.

The problem with software development is that the best design is only obvious after the first design has been built.

Don't have environments https://alexgaynor.net/2016/jan/19/dont-have-environments/ (using PRs as last step before production, without privileging a preprod environment)

Some things that might help you make better software https://www.drmaciver.com/2016/10/some-things-that-might-help-you-write-better-software/ (no panacea, but sound advice)

Considering how many languages compile to JS, it's a shame that stdlib docs rarely include a REPL.