miniblog.

Setting /bin/false as a login shell is not sufficient for SSH security:

Breakable Toys: a useful pattern for exploring new technologies with permission to build imperfect projects:

Superb blog post about "explicit is better than implicit". Are you seeking to make something opt-in, promoting 'syntactic salt', or arguing for local reasoning? https://boats.gitlab.io/blog/post/2017-12-27-things-explicit-is-not/

Slashdot is the only community site I've seen where you can vote comments along multiple axes (insightful, interesting, funny). It's really useful metadata: jokes are inevitable, so rather than having rules about them, let the site order comments appropriately.

Wuffs: a programming language that prevents null pointer defence, buffer overflow, and arithmetic overflow, all at compile time! Designed for file format parsing.

Wonderful quote on https://web.archive.org/web/20120621062217/http://web.cecs.pdx.edu/~kirkenda/joy84.html (the history of vi) describing how tools improved on the usability of ed.

Compile-to-web PLs come in three flavours (increasing difficulty of adoption): JS-like: different syntax/semantics, existing JS tooling: Typescript, CS Separate world: builds own toolchain: Elm, Purescript Separate platform: includes own runtime/needs FFI: ScalaJS, Pyjamas

Overheated chips in the Apple III could end up dislodged. Apple's recommendation? Drop it two inches!

Racket has a remarkable number of aliases. '() and empty are equivalent, as are #t, #true and true. YAML is similar, but it's unusual for a programming language.

Capacitor Plague: a remarkable period of very high capacitor failure rates during the '00s. It's believed to be a result of manufacturers trying to steal electrolyte formula but missing important parts.

I'm increasingly persuaded that languages implementations should not expose an AST and probably shouldn't have it in the stdlib either. Static analysis tools need a richer data structure, and want to parse different language versions. It's better as a standalone library.

Happy Christmas!

Formal Verification: The Gap Between Perfect Code and Reality https://raywang.tech/2017/12/20/Formal-Verification:-The-Gap-between-Perfect-Code-and-Reality/ Good critique of how formal verification techniques can be extremely heavyweight. Also discusses (rather pessimistically) how it's important to know which parts are not verified

'S-expressions are a bad idea' by notable lisper Dave Moon: https://cl.cddddr.org/index.cgi?Essay:S-expressions are a bad idea&l= Argues in favour of richer data types for syntax rather than lists and symbols. It's true: propagating positions when your macro is incorrectly used is difficult!

Great excerpt from the Realm of Racket (currently available in the Humble Bundle!) on how OO PLs lead to Scheme. This quote is from page 27.