D is for Data Science

[Chris via Digitalmars-d-announce]

On Friday, 28 November 2014 at 12:06:06 UTC, Chris wrote:
> On Tuesday, 25 November 2014 at 13:24:04 UTC, ketmar via 
> Digitalmars-d-announce wrote:
>> On Mon, 24 Nov 2014 17:10:25 -0800
>> Walter Bright via Digitalmars-d-announce
>> <digitalmars-d-announce at puremagic.com> wrote:
>>
>>> I know it's a tough call. But I do see these sorts of 
>>> comments regularly, and it is a fact that there are too many 
>>> D libraries gone to seed that won't compile anymore, and that 
>>> makes us look bad.
>> but D wins in overall. being one of the architects in my 
>> bussiness i
>> was eagerly pushing D as our main development language. it's 
>> good that
>> this thing (and some other too) happens before i succeeded. 
>> now we keep
>> going with C++, as it fscks safety too, fscks principle of 
>> least
>> astonishment, almost never fixes inconsistencies, but it has 
>> alot more
>> libraries and i can hire alot more programmers with it. i'm 
>> still using
>> D as a language for my hobbyst throw-away projects though, and 
>> D is
>> great for such things. D wins, 'cause i *almost* stopped 
>> ranting (not
>> only in this NG) and just accepting it as is. well, almost as 
>> is, i'm
>> applying alot of patches over vanilla D. this, of course, 
>> makes my code
>> incompatible with every other D compiler out here, but luckily 
>> this is
>> not a concern anymore.
>
> "just accepting it as is" - Well, there's no need to do that. 
> If there are issues, you're free to comment on them, make a 
> feature request and/or fix them yourself. Everybody accepts any 
> language "as is" as long as it's a mainstream language, 
> regardless of any shortcomings or major annoyances. Your 
> comment proves just that.
>
> Just this week I was working on new software and I'm still 
> amazed at how many options I have in D (and I keep discovering 
> new options). D is always compared to C++ in terms of 
> performance and libraries. Sure, there are more libraries (and 
> by extension programmers) out there for C++. Performance might 
> be better or worse, depending on the library and the 
> programmer. However, The sheer abundance of options and 
> modeling power in D is one of the reasons I stick with D. I 
> deal with problems concerning language processing (grammar, 
> rules etc.), i.e. mapping the human mind to a machine, and D 
> always gives me a way to model complex and intricate systems. 
> Sometimes I look at the code and think "How would I have 
> implemented this in C, Python or Java?" I shudder and say "No 
> way!" Believe it or not, modeling power, often overlooked, is 
> one of the key features of programming languages of the future. 
> Performance can always be improved. But modeling power is hard 
> to add, if you don't have it already. Libraries, well, if you 
> have strong modeling power, you can roll your own very quickly. 
> Maybe an abundance of libraries is a sign that a language lacks 
> modeling power.

About the article, it proves two things. First, you can easily 
roll your own in D. Second, you have to know the language well to 
be able to get the most out of it without having to roll your 
own.[1] Either way, it improves your general understanding of 
programming.

[1] This includes not hesitating to ask question on D.learn.