miniblog.

I love languages that go the extra mile with their syntax errors. Here's Julia being helpful when you write 'else if' rather than 'elseif'.

The @racketlang docs are superbly cross-referenced. Even the code examples convert the function names into links!

Company dev blogs are very much underrated. Thread.

Julia's generated fns are like macros,but they have access to the type of fns! https://docs.julialang.org/en/latest/manual/metaprogramming/#Generated-functions-1 (Other PLs call this staged fns AIUI)

Apparently the 140 character limit on tweets is only enforced clientside! https://www.theverge.com/2017/9/27/16372666/twitter-280-characters-guide-workaround

Blogged: These Weeks in Remacs III:

Learning about vintage approaches to find improvements for modern designs. I feel there's a parallel between PL design and fashion.

The benefit of annotating commits by feature: you can measure how many bugfixes each feature needed!

Racket seems to have a critical mass of mindshare: tons of accomplished lispers building some amazing PL features. I need to use it more.

Is 'make -jN' an anachronism? Shouldn't a modern build tool choose a sensible level of parallelism without human intervention?

"a lock-free application won’t lock up even with thread scheduling decisions being made by your worst enemy" https://morestina.net/blog/749/exploring-lock-free-rust-2-atomics

Creating a circular data structure should be impossible with immutability, right? Amazingly, laziness enables this:

We're discussing a logo for Remacs! https://github.com/Wilfred/remacs/pull/360

Too many program synthesis tools assume your spec is in a decidable format. They've had incredible results, but tests are way more general.

It's less interesting to argue in favour of a type system in general. Should it be gradual, structural, nominal and/or dependent, and why?