3 ms·
It's actually surprisingly difficult to get this kind of thing right. It seems perfectly obvious in hindsight, but while you're designing, little details like t
by kstenerud 7y ago
It's actually surprisingly difficult to get this kind of thing right. It seems perfectly obvious in hindsight, but while you're designing, little details like this get missed.
When I started writing a spec for Concise Text Encoding [1], I figured it would take about 6 months to nail it down (yeah, right). Even now, 2 years later, I'm still making amendments to it because of various things I missed or got wrong. It's either something I happen to notice in one of my many, many re-reads, or something that another person notices after a few minutes reading it over, or something that comes up while I'm writing the reference implementation. Getting a spec right is a marathon affair.
[1] https://github.com/kstenerud/concise-encoding/blob/master/cte-specification.md https://github.com/kstenerud/concise-encoding/blob/master/ct...
- funcDropShadow 7y agoIt seems to be an instance of Wadler's Law. > In any language design, the total time spent discussing > a feature in this list is proportional to two raised to > the power of its position. > 0. Semantics > 1. Syntax > 2. Lexical syntax > 3. Lexical syntax of comments https://wiki.haskell.org/Wadler's_Law https://wiki.haskell.org/Wadler's_Law