Applying type-based program synthesis to code completion, leading to snippets, not just method names. Clever!
miniblog.
What if code completion / search was implemented in a graph database? A fascinating discussion by some Haskellers:
Smalltalk is sufficiently alien that it can be hard to get started. There's a really helpful intro here: https://ceronio.net/2017/07/first-steps-with-pharo-smalltalk/
Racket's docs have some lovely examples of programming by contract: https://docs.racket-lang.org/guide/contracts-examples.html
Conditional syntax in @pharoproject still feels rather alien to me. Reviews welcome! Copy-pasteable version:
A wonderful property of lisps is that you can usually parse unfinished code into raw sexps. This is really handy when building editor tools.
You think you've seen every variant of code highlighting, then you see this fun Atom style!
Coverage percentage is misleading, as 60% coverage has a different meaning for 800 LOC vs 80 KLOC. Perhaps uncovered line count is better?
Many org-mode tutorials have a steep learning curve or assume a workflow. This great talk starts with the basics: https://youtu.be/fgizHHd7nOo
Tokei, a Rust LOC counting tool, now offers badges! You can show off how little/how much code is there.
jwz's criticisms of Java: https://www.jwz.org/doc/java.html (several no longer apply, and it's interesting to see how much he cares about object size)
Dash.el has destructuring -lambda, which is hugely underrated. (-map (-lambda ((fst snd)) snd) pairs) is a lovely way of extracting items!
The LaTeX fetish (Or: Don’t write in LaTeX! It’s just for typesetting) https://www.danielallington.net/2016/09/the-latex-fetish/ (always good to ask: 'why do we use this?')
Measuring code completion usage by devs https://www.cl.cam.ac.uk/~mcm79/pdf/2015-PPIG.pdf (~40% of completions are unhelpful, lack of completion is often useful!)
Showing 3,361-3,375 of 7,584 posts