miniblog.

How do you randomly sample from non-uniform distributions? This is a common problem when computing random numbers. TIL about rejection sampling, where you randomly sample within an area that includes the distribution, until you hit a value inside:

Here's an IDE feature I think should exist: A *random* file finder! Rather than doing a range of static analyses (typical lint issue count, typical number of functions per file), just randomly sample! It's less overwhelming and prevents recent feature bias to code cleanliness.

Cute idea (in a rather dramatic blog post): randomly generate strings for PL keywords, and apply the same transformation to your source code. This prevents code injection: https://blog.polyverse.io/introducing-polyscripting-the-beginning-of-the-end-of-code-injection-fe0c99d6f199 (I suppose you could brute force, and they don't say if it breaks eval.)