GCC 4.6

[jasonw]

dsimcha Wrote:

> On 3/28/2011 9:54 PM, jasonw wrote:
> > Listen kid, you're some biology student, right? You're just coding for fun. And more importantly, you haven't participated in any long term real world systems programming projects. This kind of work experience doesn't give you the competence to evaluate the knowledge and work of people with tens of years of programming experience under their belt.
> >
> > You might be terribly smart, but you're missing the point. Can you see what we are building here? A whole language ecosystem. Andrei has done great work by attracting competent CS persons in to the community.
> 
> While I think some good points were raised here, I find the implication 
> that biologists and generally non-CS people can't do first rate 
> programming mildly offensive.  Formal education in CS helps especially 
> when doing CS research, but it's not a requirement for being a "real" 
> programmer.  I'm a biomedical engineering student and primarily write 
> research and hobby code, not industrial code.  Walter's degree is in 
> mechanical engineering and he's one of the best programmers I can think 
> of.  Heck, even Andrei didn't have a formal degree in CS until recently. 
>   (His undergrad, IIRC, is in electrical engineering.)

Didn't want to give that kind of impression. I think "computer engineering" graduates often have a much better view of the software industry than pure computer science graduates. Non-CS people can also be great developers, no doubt.

I wanted to point out that bearophile tries to act like a main architect of the D language. I'm overall impressed how well he handles the theoretical side, but the price is that the ideas are always presented like from an ivory tower. Concrete compiler / library patches and better relevance to real "simple" everyday problems would help a lot more. I'll mention one concrete example: typestates. While I think this idea is interesting, I really can't trust bearophile if I want to know if it's a good feature to have and testing it would require studying some small unpopular language (Rust). I don't think the main branch of D is a suitable target for experimenting with this kind of features. If he wants to use them, he could fork D and come back with concrete results.

> 
> That said, I think bearophile's posts are well-intentioned.  The problem 
> is that the signal-to-noise is terrible.  What D needs now is bug fixing 
> of what's already there and solid implementations of basic stuff like 
> database APIs, better garbage collection, IDEs, etc.  Bringing up the 
> latest cool idea is fine if you've also got an implementation or it's 
> exceptionally well thought out and solves a severe, pressing problem. 
> The constant bombardment with ideas to solve minor or niche problems, 
> with no implementation and no intention of creating an implementation, 
> is more distracting than useful.

I fully agree with this, but just wanted to bring this up because in my opinion he is wasting a lot of time trying to emphasize things which have been noted, but are on the bottom of the current priority list. I don't want to discourage him from posting, but he should also consider the reactions of the audience a bit.