June 09, 2012
I do believe D is my new favorite language. I'm trying to figure out just how far you can take D... here's my question from StackOverflow:

http://stackoverflow.com/questions/10962541/is-d-powerful-enough-for-these-features

"For the longest time I wanted to design a programming language that married extensibility with efficiency. I recently rediscovered D and I am wondering if D 2.0 is pretty much the language I wanted to make myself. What I love most is the potential of metaprogramming; in theory, could D's traits system enable the following features at compile time?"

"1. Run-time reflection: Are the compile-time reflection features sufficient to build a run-time reflection system a la Java/.NET?"

"2. Code conversion: Using a metaprogram, create C#/C++/etc. versions of your D program every time you compile it (bonus point if doc comments can be propagated)."

"3. Traits. I don't mean the metaprogramming traits built into D, I mean object-oriented traits for class composition. A D program would indicate a set of traits to compose, and a metaprogram would compose them."

"4. Unit inference engine: Given some notation for optionally indicating units, e.g. unit(value), could a D metaprogram examine the following code, infer the correct units, and issue an error message on the last line? (I wrote such a thing for boo so I can assure you this is possible in general, program-wide):"
June 09, 2012
Oops, forgot the code.

> "4. Unit inference engine: Given some notation for optionally indicating units, e.g. unit(value), could a D metaprogram examine the following code, infer the correct units, and issue an error message on the last line?

auto mass = kg(2.0);
auto accel = 1.0;
auto force = mass*accel;
accel += metresPerSecondSquared(9.81); // units of 'force' and 'accel' now known
force += pounds(3.0);                  // unit mismatch detected

P.S. And can D output 'pre-processed' source, i.e. the code after metaprograms have executed?
June 10, 2012
Am Sat, 09 Jun 2012 19:07:17 +0200
schrieb "David Piepgrass" <qwertie256@gmail.com>:

> Oops, forgot the code.
> 
> > "4. Unit inference engine: Given some notation for optionally indicating units, e.g. unit(value), could a D metaprogram examine the following code, infer the correct units, and issue an error message on the last line?
> 
> auto mass = kg(2.0);
> auto accel = 1.0;
> auto force = mass*accel;
> accel += metresPerSecondSquared(9.81); // units of 'force' and
> 'accel' now known
> force += pounds(3.0);                  // unit mismatch detected

There are some implementations (I think 2 or 3 have been announced on the newsgroup) which do exactly that. Take a look at this one, for example: https://github.com/klickverbot/phobos/blob/units/std/units.d http://klickverbot.at/code/units/std_units.html http://klickverbot.at/code/units/std_si.html

There was some talk about including something like that in the standard library, but that discussion went nowhere.


> P.S. And can D output 'pre-processed' source, i.e. the code after metaprograms have executed?

There's a patch for dmd which does that, but it hasn't
been merged yet.
https://github.com/D-Programming-Language/dmd/pull/426
June 13, 2012
On Saturday, 9 June 2012 at 17:07:19 UTC, David Piepgrass wrote:
> auto mass = kg(2.0);
> auto accel = 1.0;
> auto force = mass*accel;
> accel += metresPerSecondSquared(9.81); // units of 'force' and 'accel' now known
> force += pounds(3.0);                  // unit mismatch detected

accel will be double. I guess you want bidirectional type inference. I think it was done for python, but resulted in too big compilation times (the times are ok for a helloworld).
Top | Discussion index | About this forum | D home