The D programming language - an interview with the author

[Gothi[c]]

[ Invalid Attachment ]

Recently featured on the OSNews blog was an article on compuworld with the author of the D programming language, Walter Bright.

D takes C++ and incorporates elements from more modern programming languages such as ruby,python,java,etc...
I find this particularly interesting because D remains a true systems programming language, with many of the original concepts that make me stick to C++ in it.

It looks like D has come a long way since I last tried it, and they are getting ready for their 2.0 release (alpha has been released now).

D is not without it's problems, for example, it's standard library has been forked and is not compatible with the original stdlib. Bright addresses this and other criticisms on the language in the article above.

Very interesting read!

[mouser]

The interview has a link to some video presentations you can watch which are nice too.

Also relevant to this post are these intervew with Stroustrup on c++ and c++0x:
https://www.donation...dex.php?topic=9559.0 and https://www.donation...dex.php?topic=6418.0

D has a lot of nice things going for it and if you read my comments on the stroustrup threads you'll see i'm not enthusiastic about c++0x.  But i still don't "love" D, so i'm still waiting for another evolution from C++ to capture my heart..

[mouser]

There's also a new book available on D, which seems very short and isn't getting great reviews, but is cheap:



http://www.amazon.co...ativeASIN=1590599608

[Gothi[c]]

There's also a new book available on D, which seems very short and isn't getting great reviews, but is cheap

-mouser (July 26, 2008, 11:03 AM)

Cool! This is a book from 2007, so I guess it's fairly recent, which is important because D has changed a lot over the years.

[f0dder]

There's some cute ideas in D, but I still see it as kinda a niche language. I don't need garbage collection (I prefer object-lifetime destruction & knowing when I get a performance hit), and with BOOST's for_each macro I find my needs pretty much covered.

So I don't really know about D. There's things I'd definitely like to see in (core) C++, but I wouldn't jump wagon and use D since C++ is available just about everywhere (and with decent optimizing compilers), whereas D isn't.

[Ehtyar]

Forgive me for resurrecting an old thread - I found it in the "Best of 2008". I wouldn't have replied except I didn't check the date, and now it seems rather silly to discard my post simply due to that. I wonder, in the last six months has anyone's opinion changed?

The thing that really turned me off D was the fiasco with the "standard library", otherwise known as a mishmash of Tango and it's piddly little forerunner, Phobos. Once the two are completely integrated, I'd be interested in using D, but from where I'm sitting at the moment, there really isn't enough incentive to move from C++ which is far more widely used, supported and exemplified.

Ehtyar.

[f0dder]

Imho, with C++0x a-coming and TR1 already available, the incentive for D is smaller. Sure, it'd be nice to have a few of D's features like modules (rather than the C/C++ include-file/library mess), but meh - C++ is industry standard and has some pretty great compilers.

[Gothi[c]]

* Gothi[c] waits for mouser's rant on C++0x in 5...4...3....

[mouser]

My main ranting against C++0x can be found here: https://www.donation...59.msg71904#msg71904

Summary:

• It's still my primary programming language, and I still love it, but you simply cannot "fix" C++ by adding stuff, which is the ground rules that C++0x is operating under.  It's simply not feasible, and the result is ugly.
• In truth, the C++0x debacle is going to have some very useful side effects.  Their are some brilliant language designers working on C++0x issues, and churning out interesting, detailed, thoughtfull, argued positions.  I suspect that many of the ideas that come out of C++0x arguments will find their way into younger, more consistent and more elegant languages.

I actually suspect that the introduction of C++0x is going to ironically hasten the fall of C++.  The momentum to stick with a language is very strong.. But when you force people to change, you've provided them with an incentive to switch to another language rather than adopt a new version of C++.

--

I've read the D book and have decided D is not worth my switching to it from C++.  Thus, I am still waiting for a new language.

[mouser]

ps i keep a little notebook of ideas for what i would consider my ultimate dream language.  if anyone here ever becomes a serious language designer with a few years to spend writing the next new language, drop me a note 

[f0dder]

I don't really see why C++0x should "speed up the downfall". There currently aren't any other languages out there that are able to replace C++ fully, because of the sheer flexibility of the language.

0x adds important stuff like lambdas (queer syntax perhaps, but it's finally there) and for_each. And gives the "auto" keyword some meaning in a "ohgod I've missed this when doing template programming" way. And getting a shared_ptr in the standard library is nice, even though it's already available with boost.

Yes, C++ has problems. The standard library kinda sucks, it's very hard writing a decent parser for it (which also means getting proper intellisense in an IDE is a major pain), and (because the standard library kinda sucks) you can blow your leg off if you aren't careful. And for a lot of applications on today's hardware you can get away with using a language like C# or Java just fine.

There still isn't anything beating C++, though.

[Eóin]

I'm not sure I agree with you on C++ mouser. Having read (and tried to digest) Imperfect C++ by Matthew Wilson, author of the STLSoft libraries, I've had all my faith in the language restored. The book is a 500+ page tomb of everything that's wrong (well imperfect) with C++ and very applicable solutions to each one.

Honestly as it stands I can only see C++0x (which I believe is just an informal name while under standardization) improving the situation. Maybe just me, a more detailed debate should be moved to it's own space I suspect.

[mouser]

Imperfect C++ and stroustrup's Design and Evolution of C++ are both good books -- I enjoyed them both thoroughly.  Doesn't change my pessimistic view though.  I also definitely can recommend the slim volume on D I mentioned above.

[Ehtyar]

If C++0x were to be released today, it would be named C++09 in a similar fashion to the 1999 revision of C being named C99. the 0x is simply a placeholder for the abbreviated year of release.

Ehtyar.

[Gothi[c]]

I guess they better hurry then, or they'll have to rename it to C++1x

[cranioscopical]

It might be too late!

0xzen

0/O who knew?

[Ehtyar]

I guess they better hurry then, or they'll have to rename it to C++1x

-Gothi[c] (January 31, 2009, 09:26 PM)

That's a good point. Are we to assume that C++0x will be finalized this year?...I think not

Ehtyar.