On 02/09/2011 11:43 PM, Nick Sabalausky wrote:
Although I do like the inverse approach that D ended up taking: Don't bother
with simplicity/orthogonality at first, just get important features in.
*Then*  refactor the internals to shuffle the complexity into the std lib and
simplify the core language. (And, of course, use lowerings whenever
appropriate.)

I love that. I often program that way, at least up to a certain size/complexity.
I wouldn't even dare designing anything as big and complex as a PL using that approach. It's a question of amount of information and level of abstraction one is able to master, I guess. (I know from experience that my capacity of abstraction is not very powerful, esp compared to top-programmers). Also, I would tend to conclude from (what I know of) D's past & present development history that it a good approach: leads to an eternal beta state (or even alpha)?

Funny though, right after I posted "You'd think that things like JS...", the
obvious devil's-advocate counter-argument occurred to me: "You'd think that
things like C++ and Algol would have taught people that complex languages
are a stupid way to go." Oh well.

Yes, orthogonality, simplicity, lead to clarity. Which means people (with some rational skill) can far more easily grasp the system. But practical usability often requires breaking the nice, seducing, theoretical schemes. I guess. Or requires adding level of complexity to restore consistency after introducing practical features, notions, dictinctions. Veeery difficult. But a language designed from the start on with only praticality in mind, with consistency not considered a design criterion, maybe does not even have a chance to become a beautiful, lovable and fun tool. C++. Except for toy cases,

denis
--
_________________
vita es estrany
spir.wikidot.com

Reply via email to