miniblog.

On a belated note, it's really nice how fast the @melpa_emacs web UI is. I like it for those rare cases when I don't have an Emacs instance.

Learn today that Emacs used to provide a "barebin" executable without libraries (but disk space is cheap today!) http://t.co/MwVialmv2g

One thing data science has taught me is that I measure too rarely, collect too few samples and graph too rarely.

"So the whole argument that people should parallelise their code is fundamentally flawed." -- Strong words from Linus http://t.co/PFYXCL4H5J

It's rather disappointing that CPU benchmarks can report different results by just changing the vendor string: http://t.co/rMYLn8IN5K

Pragmatic reasons why you can't (in all circumstances) just add floats to a language and simulate integers:

Modern x86-64 CPUs chips have become radically more complex than the original x86 designs: http://t.co/D7V9IarSug fascinating article

Io is amazingly elegant language. The syntax is tiny yet readable.

I learn recently that JSHint is deprecating style warnings in favour of JSCS. The node.js community has amazing modularity.

Bash shortcuts are wonderful, but I worry about typos. One day I'm going to write `sudo !1` instead of `sudo !!` and cry.

I really like how clang reports type warnings that include type aliases. It's already helped me fix a bug.

Super nifty web-browsable C++ source code with cross references: http://t.co/AHlOzNNeEQ -- why don't we have this for more languages?

SPEC benchmarks depend on undocumented C behaviour: http://t.co/9zkybqwM9N -- compiler benchmarking takes real language lawyering.

I saw a company today describe itself as using MangoDB. I really hope that was a typo:

I'm envious of gofmt, which formats Go code in a canonical style. TIL clang-format supports several languages, so I'm giving it a try for C.