3 ms·
It is true that there is at best very little evidence that code gets better. (Some studies found very, very small positive effects, others found none, some wen
by mpweiher 8y ago
It is true that there is at best very little evidence that code gets better. (Some studies found very, very small positive effects, others found none, some went the other way).
However, the documentation effect is well-supported, and also something I find to be true anecdotally: if you have a library or code-base that you need to figure out, the types really help.
On the other hand, it is unclear whether very large software systems such as the web would be possible as statically typed systems.
- PaulHoule 8y ago"The web" is an interesting one. I think web browsers are overwhelming implemented in systems languages, but people often write other clients in scripting languages. Javascript is the dominant client-side scripting language but many people like Typescript, GWT, Dart and things like that. You certainly can write dynamic web sites in a systems language, but almost always people use a scripting language like Perl, PHP, Python, Cold Fusion, etc. Java may be the exception that "proves the rule" but Java tends to have escape hatches for compile-time binding such as Spring.
- mpweiher 8y agoYou can certainly write specific sites using whatever you want. However, the web as a whole is and has to be dynamically typed.