www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Re: Slides from my ACCU Silicon Valley talk

reply bearophile <bearophileHUGS lycos.com> writes:
Andrei:

 http://erdani.com/tdpl/2010-12-08-ACCU.pdf



I have a small question. At page 34 of the slides it says:


 - Built-in complex types are being replaced by library types



Are complex types totally replaced, or is the complex literals syntax (like
10+10i) kept? Keeping those literals may be good.

Bye,
bearophile
Dec 13 2010
parent reply Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 12/13/10 6:11 AM, bearophile wrote:

 Andrei:

 http://erdani.com/tdpl/2010-12-08-ACCU.pdf



 I have a small question. At page 34 of the slides it says:


 - Built-in complex types are being replaced by library types



 Are complex types totally replaced, or is the complex literals syntax (like
10+10i) kept? Keeping those literals may be good.



Walter wants to keep complex literals. I strongly believe they are 
completely useless.

Andrei
Dec 13 2010
next sibling parent bearophile <bearophileHUGS lycos.com> writes:
Andrei:


Walter wants to keep complex literals. I strongly believe they are completely
useless.<



Thank you for your answer. So far I have used D complex numbers only two times,
so I don't have to use them often, so I don't feel a strong need for them. But
I like that people that use them often will find a handy syntax for those
literals.

Generally I like good literals, it's one of the differences between D and dumb
languages. For example I'd like DMD to support multi-precision integer literals
too:
bigint x = 100_000_000_000_000_000_000_000_000_000_000_000_000;

We may write down a list of all the advantages and disadvantages of complex
literals (and I'd like someone that writes code that processes complex values
to help in the writing of this list), and then try to decide rationally to keep
them or remove them. I don't have strong feelings about this topic, so I will
have no problems with both results.

Bye,
bearophile
Dec 13 2010
prev sibling next sibling parent reply "Lars T. Kyllingstad" <public kyllingen.NOSPAMnet> writes:
On Mon, 13 Dec 2010 09:29:15 -0600, Andrei Alexandrescu wrote:


 On 12/13/10 6:11 AM, bearophile wrote:

 Andrei:

 http://erdani.com/tdpl/2010-12-08-ACCU.pdf



 I have a small question. At page 34 of the slides it says:


 - Built-in complex types are being replaced by library types



 Are complex types totally replaced, or is the complex literals syntax
 (like 10+10i) kept? Keeping those literals may be good.



 Walter wants to keep complex literals. I strongly believe they are
 completely useless.



I agree with this.  It would be interesting to know how often people 
actually write complex literals.  I suspect it is *very* rare.

And how would it work, anyway?  Should we be required to import 
std.complex to use complex literals?

In my opinion, when the built-in complex types are deprecated, the 
literals should go as well.

-Lars
Dec 14 2010
next sibling parent reply Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 12/14/10 2:56 AM, Lars T. Kyllingstad wrote:

 On Mon, 13 Dec 2010 09:29:15 -0600, Andrei Alexandrescu wrote:


 On 12/13/10 6:11 AM, bearophile wrote:

 Andrei:

 http://erdani.com/tdpl/2010-12-08-ACCU.pdf



 I have a small question. At page 34 of the slides it says:


 - Built-in complex types are being replaced by library types



 Are complex types totally replaced, or is the complex literals syntax
 (like 10+10i) kept? Keeping those literals may be good.



 Walter wants to keep complex literals. I strongly believe they are
 completely useless.



 I agree with this.  It would be interesting to know how often people
 actually write complex literals.  I suspect it is *very* rare.

 And how would it work, anyway?  Should we be required to import
 std.complex to use complex literals?

 In my opinion, when the built-in complex types are deprecated, the
 literals should go as well.



I asked people at a seminar during a recess: "How many remarkable 
complex constants do you know?" After a long pause, one person replied: 
"1 + 0i".

That brought down the house.


Andrei
Dec 14 2010
parent reply bearophile <bearophileHUGS lycos.com> writes:
Andrei:


 I asked people at a seminar during a recess: "How many remarkable 
 complex constants do you know?" After a long pause, one person replied: 
 "1 + 0i".



Instead of asking that question, I suggest to ask people that write code that
uses complex numbers if they like the idea of complex literals. And I suggest
to analyse some Fortran/C code that uses complex numbers and count how often
literals may be useful in the code, and how much they may improve the
readability of the code.

Bye,
bearophile
Dec 14 2010
parent reply Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 12/14/10 3:24 AM, bearophile wrote:

 Andrei:


 I asked people at a seminar during a recess: "How many remarkable
 complex constants do you know?" After a long pause, one person
 replied: "1 + 0i".



 Instead of asking that question, I suggest to ask people that write
 code that uses complex numbers if they like the idea of complex
 literals.



So you mean I should ask _that_ question but _other_ people.


 And I suggest to analyse some Fortran/C code that uses
 complex numbers and count how often literals may be useful in the
 code, and how much they may improve the readability of the code.



Magic constants have issues - be they strings, integrals, or doubles. I 
have trouble thinking that complex literals are somehow exempt.


Andrei
Dec 14 2010
parent bearophile <bearophileHUGS lycos.com> writes:
Andrei:


 So you mean I should ask _that_ question but _other_ people.



I don't understand what you mean.
There right questions need to be asked to the right people. I think most
C/C++/D/Java programmers don't use complex numbers, so asking to generic
programmers about complex numbers in their code may be useful just to know how
often complex numbers are used in programs, but it can't tell you much about
how good complex literal are.
If you want to design complex numbers in D, you have to ask to people that
surely have written more than 100 lines of code that uses complex numbers (in
Python too there are complex literals, so some Python programmer too may be
useful for such questions).



 Magic constants have issues - be they strings, integrals, or doubles. I 
 have trouble thinking that complex literals are somehow exempt.



I partially agree. They need to be used with care (and defined only once in the
code). On the other hand removing all integral literals from my D programs
isn't going to improve them :-)

Bye,
bearophile
Dec 14 2010
prev sibling parent reply Don <nospam nospam.com> writes:
Lars T. Kyllingstad wrote:

 On Mon, 13 Dec 2010 09:29:15 -0600, Andrei Alexandrescu wrote:
 

 On 12/13/10 6:11 AM, bearophile wrote:

 Andrei:

 http://erdani.com/tdpl/2010-12-08-ACCU.pdf





 - Built-in complex types are being replaced by library types



 (like 10+10i) kept? Keeping those literals may be good.



 completely useless.



 I agree with this.  It would be interesting to know how often people 
 actually write complex literals.  I suspect it is *very* rare.
 
 And how would it work, anyway?  Should we be required to import 
 std.complex to use complex literals?
 
 In my opinion, when the built-in complex types are deprecated, the 
 literals should go as well.
 
 -Lars



Agreed. I've just looked through some code that I thought used them 
extensively, but found only two complex literals: 1i (dozens of 
instances) and 2i (one instance).
Dec 14 2010
parent reply "Alex_Dovhal" <alex_dovhal yahoo.com> writes:
"Don" <nospam nospam.com> wrote:

 Agreed. I've just looked through some code that I thought used them 
 extensively, but found only two complex literals: 1i (dozens of instances) 
 and 2i (one instance).



So if D removes complex literals what the proposed name  for Imaginary One 
would be?
_i  ,   i_  ,   _i_,  _I  ,   I  ,   M_I   ??
 i    j   looks good to me, I'd like even having them both.

Compare code:
0.998f + 2.72e-2fi
0.998f +  i*2.72e-2f
0.998f + 2.72e-2f*  j
Dec 14 2010
next sibling parent "Alex_Dovhal" <alex_dovhal yahoo.com> writes:
Oh sorry those  i and  j also would be build in. Ignore this.
Dec 14 2010
prev sibling parent reply Don <nospam nospam.com> writes:
Alex_Dovhal wrote:

 "Don" <nospam nospam.com> wrote:

 Agreed. I've just looked through some code that I thought used them 
 extensively, but found only two complex literals: 1i (dozens of instances) 
 and 2i (one instance).



 So if D removes complex literals what the proposed name  for Imaginary One 
 would be?
 _i  ,   i_  ,   _i_,  _I  ,   I  ,   M_I   ??
  i    j   looks good to me, I'd like even having them both.
 
 Compare code:
 0.998f + 2.72e-2fi
 0.998f +  i*2.72e-2f
 0.998f + 2.72e-2f*  j



Complex(9.998f, 2.72e-2f)
Dec 15 2010
next sibling parent Ellery Newcomer <ellery-newcomer utulsa.edu> writes:
On 12/15/2010 08:34 AM, Lars T. Kyllingstad wrote:

 And if someone *really* wants to put the "i" in there, they can just
 define it in their own app:

    enum i = Complex!double(0, 1);
    auto z = 0.998 + 2.72e-2*i;

 -Lars



or Complex!"0.998 + 2.72e-2i"
Dec 15 2010
prev sibling parent "Alex_Dovhal" <alex_dovhal yahoo.com> writes:
"Lars T. Kyllingstad" <public kyllingen.NOSPAMnet> wrote:

 And if someone *really* wants to put the "i" in there, they can just
 define it in their own app:

  enum i = Complex!double(0, 1);
  auto z = 0.998 + 2.72e-2*i;

 -Lars



Looks very nice.
Dec 15 2010
prev sibling next sibling parent spir <denis.spir gmail.com> writes:
On Tue, 14 Dec 2010 05:15:05 -0500
bearophile <bearophileHUGS lycos.com> wrote:


 Magic constants have issues - be they strings, integrals, or doubles. I=







 have trouble thinking that complex literals are somehow exempt. =20



 I partially agree. They need to be used with care (and defined only once =



ograms isn't going to improve them :-)

What about (char &) string literals? Hello, World! in D:

void main () {
    string message =3D [
        0x48,
        0x65,
        0x6c,
        0x6c,
        0x6f,
        0x2c,
        0x20,
        0x57,
        0x6f,
        0x72,
        0x6c,
        0x64,
        0x21,
    ];
    writeln(message);
}

And be happy that int's auto-cast to char's! Else, it would be:

void main () {
    string message =3D [
        cast(char)0x48,
        cast(char)0x65,
        cast(char)0x6c,
        cast(char)0x6c,
        cast(char)0x6f,
        cast(char)0x2c,
        cast(char)0x20,
        cast(char)0x57,
        cast(char)0x6f,
        cast(char)0x72,
        cast(char)0x6c,
        cast(char)0x64,
        cast(char)0x21,
    ];
    writeln(message);
}

(works!)
(Should we put this version on Wikipedia's Hello, World! page http://en.wik=
ipedia.org/wiki/Hello_world_program_examples ?)


Denis
-- -- -- -- -- -- --
vit esse estrany =E2=98=A3

spir.wikidot.com
Dec 14 2010
prev sibling next sibling parent so <so so.do> writes:
On Tue, 14 Dec 2010 10:56:18 +0200, Lars T. Kyllingstad  
<public kyllingen.nospamnet> wrote:


 On Mon, 13 Dec 2010 09:29:15 -0600, Andrei Alexandrescu wrote:


 On 12/13/10 6:11 AM, bearophile wrote:

 Andrei:

 http://erdani.com/tdpl/2010-12-08-ACCU.pdf



 I have a small question. At page 34 of the slides it says:


 - Built-in complex types are being replaced by library types



 Are complex types totally replaced, or is the complex literals syntax
 (like 10+10i) kept? Keeping those literals may be good.



 Walter wants to keep complex literals. I strongly believe they are
 completely useless.



 I agree with this.  It would be interesting to know how often people
 actually write complex literals.  I suspect it is *very* rare.

 And how would it work, anyway?  Should we be required to import
 std.complex to use complex literals?

 In my opinion, when the built-in complex types are deprecated, the
 literals should go as well.

 -Lars



My first and only proposal to this newsgroup was about literals.
C like literals already come with a few serious issues if you use them in  
a template heavy code.
For example a C++ template (which meant to be a generic solution) full of  
"casts" yes casts!, and this is not an unusual practice
The replies i got made me think that D solves some of these.

For the topic.
Having complex literals cool, but i can live without them!

-- 
Using Opera's revolutionary email client: http://www.opera.com/mail/
Dec 14 2010
prev sibling next sibling parent "Simen kjaeraas" <simen.kjaras gmail.com> writes:
Alex_Dovhal <alex_dovhal yahoo.com> wrote:


 "Don" <nospam nospam.com> wrote:

 Agreed. I've just looked through some code that I thought used them
 extensively, but found only two complex literals: 1i (dozens of  =







 instances)
 and 2i (one instance).



 So if D removes complex literals what the proposed name  for Imaginary=





 One
 would be?
 _i  ,   i_  ,   _i_,  _I  ,   I  ,   M_I   ??
  i    j   looks good to me, I'd like even having them both.

 Compare code:
 0.998f + 2.72e-2fi
 0.998f +  i*2.72e-2f
 0.998f + 2.72e-2f*  j



With D supporting Unicode, it should clearly be =E2=84=91.


-- =

Simen
Dec 14 2010
prev sibling next sibling parent "Lars T. Kyllingstad" <public kyllingen.NOSPAMnet> writes:
On Wed, 15 Dec 2010 15:20:01 +0100, Don wrote:


 Alex_Dovhal wrote:

 "Don" <nospam nospam.com> wrote:

 Agreed. I've just looked through some code that I thought used them
 extensively, but found only two complex literals: 1i (dozens of
 instances) and 2i (one instance).



 So if D removes complex literals what the proposed name  for Imaginary
 One would be?
 _i  ,   i_  ,   _i_,  _I  ,   I  ,   M_I   ??  i    j   looks good to
 me, I'd like even having them both.
 
 Compare code:
 0.998f + 2.72e-2fi
 0.998f +  i*2.72e-2f
 0.998f + 2.72e-2f*  j



 Complex(9.998f, 2.72e-2f)



And if someone *really* wants to put the "i" in there, they can just 
define it in their own app:

  enum i = Complex!double(0, 1);
  auto z = 0.998 + 2.72e-2*i;

-Lars
Dec 15 2010
prev sibling parent spir <denis.spir gmail.com> writes:
On Wed, 15 Dec 2010 14:34:47 +0000 (UTC)
"Lars T. Kyllingstad" <public kyllingen.NOSPAMnet> wrote:


 And if someone *really* wants to put the "i" in there, they can just=20
 define it in their own app:
=20
   enum i =3D Complex!double(0, 1);
   auto z =3D 0.998 + 2.72e-2*i;



+++

Denis
-- -- -- -- -- -- --
vit esse estrany =E2=98=A3

spir.wikidot.com
Dec 15 2010