miniblog.

Slightly scary to learn that C++ can instantiate objects for you, implicitly: http://t.co/UMFCbtsuWN (unless you use the 'explicit' keyword)

"Combines the speed of BASIC with the ease of INTERCAL and the readability of an IOCCC entry!" http://t.co/CjwJ1NOdK9

Whilst C++ is a curly-brace language, it's really quite different syntactically to C/JS/Java (even more in C++11). Seeking beginner's mind.

A single-instruction set computer can of course have a simpler design! Why would you want anything else? http://t.co/VdE0eZFD6u (spoof)

Fantastic discussion of register allocation limitations on HN:

Most of programming productivity is finding ways to reduce the iteration cycle. Live coding > linting > compile warnings > tests > usage.

I can't imagine using Twitter to discuss a hot-button/highly politicised topic. It would just be setting yourself up for a torrent of abuse.

Wow, I had no idea @pharocloud existed. Very cool to see Pharo-specific application hosting!

Interesting to see a Rust raytracer compile faster than the equivalent C++: https://ruudvanasseldonk.com/2014/10/20/writing-a-path-tracer-in-rust-part-7-conclusion - that itself is a compelling advantage.

New version of pypy is out! http://t.co/QcTwH3qbfe 2.5.0 shows some impressive performance gains in an already mature project.

Learnt today that x86 has UD2, a deliberately invalid opcode for testing. It's a big instruction set.

LLVM is the first curly-braced language project I've hacked on that uses UpperCaseVariableNames. New conventions are always hard at first.

Pharo 4 will define ifTrue: on non-booleans, adding truthiness to Pharo (for better or worse): http://t.co/A7wNcGVqa9

Generational GC lands in Julia! https://github.com/JuliaLang/julia/pull/8699 I'm also flattered some of my benchmarks were useful.

"I was surprised by the power and I say that as a Lisp programmer that's used to modifying live systems." http://t.co/9ODcn5L070 (on Pharo)