www.digitalmars.com         C & C++   DMDScript  

D - How do you like result variable ?

reply "Y.Tomino" <demoonlit inter7.jp> writes:
Many functions are written as follows:

int func(...)
{
  int result;
  ...
  result = ...
  ...
  return result;
}

I feel it's needless bother.
I prefer auto-declared "result" variable like Eiffel or Delphi.

int func(...)
{
  ...
  result = ...
  ...
}

It simplifies code.

And return is one of jump statements.
Although I don't dislike a jump statement, I don't want to use it where jump
is unnecessary.
Even if result introduced, the compatibility will be kept if return will set
result before exit.

Thanks.
YT
Jan 31 2004
next sibling parent Georg Wrede <Georg_member pathlink.com> writes:
I agree. 

The shorter your functions, the more you suffer from this.
Borland has done a number of things right, and this is
one of them.


In article <bvhb5v$2uc4$1 digitaldaemon.com>, Y.Tomino says...

Many functions are written as follows:

int func(...)
{
  int result;
  ...
  result = ...
  ...
  return result;
}

I feel it's needless bother.
I prefer auto-declared "result" variable like Eiffel or Delphi.

int func(...)
{
  ...
  result = ...
  ...
}

It simplifies code.

And return is one of jump statements.
Although I don't dislike a jump statement, I don't want to use it where jump
is unnecessary.
Even if result introduced, the compatibility will be kept if return will set
result before exit.

Thanks.
YT
Jan 31 2004
prev sibling next sibling parent reply "Matthew" <matthew.hat stlsoft.dot.org> writes:
Pointless sugar.

Sorry

"Y.Tomino" <demoonlit inter7.jp> wrote in message
news:bvhb5v$2uc4$1 digitaldaemon.com...

 Many functions are written as follows:

 int func(...)
 {
   int result;
   ...
   result = ...
   ...
   return result;
 }

 I feel it's needless bother.
 I prefer auto-declared "result" variable like Eiffel or Delphi.

 int func(...)
 {
   ...
   result = ...
   ...
 }

 It simplifies code.

 And return is one of jump statements.
 Although I don't dislike a jump statement, I don't want to use it where


jump

 is unnecessary.
 Even if result introduced, the compatibility will be kept if return will


set

 result before exit.

 Thanks.
 YT
Jan 31 2004
next sibling parent reply "Y.Tomino" <demoonlit inter7.jp> writes:
"return" has two functions(return value and exit),
I think it's unnatural syntax sugar that makes redundant code.

However, as you wrote, in the language which already has return, it's
unavoidable that another method seems to pointless.
Thank you.

YT
Feb 01 2004
parent reply Ilya Minkov <minkov cs.tum.edu> writes:
Y.Tomino wrote:

 "return" has two functions(return value and exit),
 I think it's unnatural syntax sugar that makes redundant code.


It's not even "sugar"...

I think return variable syntax is also natural because space for return
variable is pre-allocated on the stack, and it gives you the feeling and
the compiler the ease to write into it directly. C/C++ must do return
optimisation for that.


 However, as you wrote, in the language which already has return, it's
 unavoidable that another method seems to pointless.
 Thank you.


I doesn't seem to be true. Delphi has 4 ways of returning a result:
- assign to function name, which is an implicit return variable, then
wait for function exit. I think it breaks when used within classes.
- assign to implicit varible Return, then wait for function exit. This
is used most often.
- return a value directly using a return statement.
- assign to one of the abovenamed imlicit return variables, then exit
function with a return statement without value, IIRC.

Delphi makes strong compile-time control to ensure that each function
returns a valid result, but Walter finds such things annoying because
they sometimes go wrong. Thus the no return assertion and this design
will probably be held - although it's not of much advantage IMHO.

-eye
Feb 05 2004
parent reply "Y.Tomino" <demoonlit inter7.jp> writes:
It's not even "sugar"...


Sorry, I said exaggeration.


I think return variable syntax is also natural because space for return
variable is pre-allocated on the stack, and it gives you the feeling and
the compiler the ease to write into it directly. C/C++ must do return
optimisation for that.


Yes.
Pascal handle the returing area as the variable, and C's statement copy a
value to the returning area.
(However, although these are same when optimized,
The method of Pascal looks natural to me.)


 - return a value directly using a return statement.


??
Delphi don't have C-style "return" statement.


 Delphi makes strong compile-time control to ensure that each function
 returns a valid result, but Walter finds such things annoying because
 they sometimes go wrong. Thus the no return assertion and this design
 will probably be held - although it's not of much advantage IMHO.


I think no problem, because D Compiler initialize local variables to the
default value.

YT
Feb 05 2004
parent Ilya Minkov <minkov cs.tum.edu> writes:
Y.Tomino wrote:

 Sorry, I said exaggeration.


You really don't have to excuse yourself. These are all opinions. I say 
tons of wrong stuff.


- return a value directly using a return statement.


 
 ??
 Delphi don't have C-style "return" statement.


Whoops! True, i just looked up. I think it's extension by GNU Pascal and 
alike. So it doesn't count.



 I think no problem, because D Compiler initialize local variables to the
 default value.


The problem is, result variable would mean that a function would have to 
have implicit return at the end. Currently, it is different, at the end 
there is an implicit debugging assertion statement, which helps guard 
against programming errors in functions which have multiple return 
points and should always return in the middle and not in the end. 
Implicit return would give up safety. I would believe Walter would be 
against, and i don't see enough merit either. Every one-liner which uses 
a result asseignment translates into a one-liner of return statement. 
For longer functions, it doesn't matter that much anyway.

What could be done, is automatically pre-declare result variable and 
make return statement without argument return it. Would save 1 1/2 lines 
on longer functions, and possibly make easier to read. But i still don't 
see much sense.

-eye
Feb 06 2004
prev sibling next sibling parent Mark T <Mark_member pathlink.com> writes:
In article <bvhdr9$19n$1 digitaldaemon.com>, Matthew says...

Pointless sugar.

Sorry


agreed - because D is meant to attract C, C++, Java programmers that are use to
a certain idiom, at this point I would find reading "result" code to be
confusing after so many years of "return"

I think Walter made this pretty clear when he laid out his design goals for D.



"Y.Tomino" <demoonlit inter7.jp> wrote in message
news:bvhb5v$2uc4$1 digitaldaemon.com...

 Many functions are written as follows:

 int func(...)
 {
   int result;
   ...
   result = ...
   ...
   return result;
 }

 I feel it's needless bother.
 I prefer auto-declared "result" variable like Eiffel or Delphi.

 int func(...)
 {
   ...
   result = ...
   ...
 }

 It simplifies code.

 And return is one of jump statements.
 Although I don't dislike a jump statement, I don't want to use it where


jump

 is unnecessary.
 Even if result introduced, the compatibility will be kept if return will


set

 result before exit.

 Thanks.
 YT
Feb 01 2004
prev sibling parent reply Roel Mathys <roel.mathys yucom.be> writes:
Matthew wrote:


 Pointless sugar.
 
 Sorry
 


pointless remark,

sorry



not speaking about this particular case, but there are lots of thingies 
in a programming language that are meant as sugar,

e.g. in D there are multiple ways of defining:
	- function pointers
	- templates
	- ...
	or delegates, and the list goes on

bye,
roel
Feb 01 2004
parent reply The one Haranguer <The_member pathlink.com> writes:
In article <bvj62m$2tvl$1 digitaldaemon.com>, Roel Mathys says...

Matthew wrote:


 Pointless sugar.
 
 Sorry
 


pointless remark,

sorry



not speaking about this particular case, but there are lots of thingies 
in a programming language that are meant as sugar,


Ah, but not _pointless_ sugar.
Feb 01 2004
parent reply "Sean L. Palmer" <palmer.sean verizon.net> writes:
One man's pointless is another man's crusade.  It's all personal opinion,
and should be taken as such.  At least it didn't take much time to read his
reply!

Sean

The one Haranguer wrote:
| In article <bvj62m$2tvl$1 digitaldaemon.com>, Roel Mathys says...
||
|| Matthew wrote:
||
||| Pointless sugar.
|||
||| Sorry
|||
||
|| pointless remark,
||
|| sorry
||
||
||
|| not speaking about this particular case, but there are lots of
|| thingies in a programming language that are meant as sugar,
||
| Ah, but not _pointless_ sugar.
Feb 01 2004
parent reply "Matthew" <matthew.hat stlsoft.dot.org> writes:
It does nothing for my ego control that my succinct, but entirely on-point,
response has spawned discussion at a current ratio of 86:3.

:)

"Sean L. Palmer" <palmer.sean verizon.net> wrote in message
news:bvjn05$ob7$1 digitaldaemon.com...

 One man's pointless is another man's crusade.  It's all personal opinion,
 and should be taken as such.  At least it didn't take much time to read


his

 reply!

 Sean

 The one Haranguer wrote:
 | In article <bvj62m$2tvl$1 digitaldaemon.com>, Roel Mathys says...
 ||
 || Matthew wrote:
 ||
 ||| Pointless sugar.
 |||
 ||| Sorry
 |||
 ||
 || pointless remark,
 ||
 || sorry
 ||
 ||
 ||
 || not speaking about this particular case, but there are lots of
 || thingies in a programming language that are meant as sugar,
 ||
 | Ah, but not _pointless_ sugar.
Feb 01 2004
next sibling parent Roel Mathys <roel.mathys yucom.be> writes:
Matthew wrote:

 It does nothing for my ego control that my succinct, but entirely on-point,
 response has spawned discussion at a current ratio of 86:3.
 
 :)
 
 "Sean L. Palmer" <palmer.sean verizon.net> wrote in message
 news:bvjn05$ob7$1 digitaldaemon.com...
 

One man's pointless is another man's crusade.  It's all personal opinion,
and should be taken as such.  At least it didn't take much time to read


 
 his
 

reply!

Sean

The one Haranguer wrote:
| In article <bvj62m$2tvl$1 digitaldaemon.com>, Roel Mathys says...
||
|| Matthew wrote:
||
||| Pointless sugar.
|||
||| Sorry
|||
||
|| pointless remark,
||
|| sorry
||
||
||
|| not speaking about this particular case, but there are lots of
|| thingies in a programming language that are meant as sugar,
||
| Ah, but not _pointless_ sugar.


 
 
 


10 points for you (on a scale of ...) :-)

see ya,
rm
Feb 01 2004
prev sibling parent reply "Sean L. Palmer" <palmer.sean verizon.net> writes:
If you want to boost that, do a one-character reply next time.  ;)

Sean

Matthew wrote:
| It does nothing for my ego control that my succinct, but entirely
| on-point, response has spawned discussion at a current ratio of 86:3.
|
| :)
Feb 01 2004
next sibling parent "Matthew" <matthew.hat stlsoft.dot.org> writes:
k

"Sean L. Palmer" <palmer.sean verizon.net> wrote in message
news:bvjs4s$11bf$1 digitaldaemon.com...

 If you want to boost that, do a one-character reply next time.  ;)

 Sean

 Matthew wrote:
 | It does nothing for my ego control that my succinct, but entirely
 | on-point, response has spawned discussion at a current ratio of 86:3.
 |
 | :)
Feb 01 2004
prev sibling parent reply The Lone Haranguer <The_member pathlink.com> writes:
In article <bvjs4s$11bf$1 digitaldaemon.com>, Sean L. Palmer says...

If you want to boost that, do a one-character reply next time.  ;)

Sean

Matthew wrote:
| It does nothing for my ego control that my succinct, but entirely
| on-point, response has spawned discussion at a current ratio of 86:3.
|
| :)


Is there a WingDing or Unicode character of "the finger"?
Feb 02 2004
parent reply "Serge K" <skarebo programmer.net> writes:
 Is there a WingDing or Unicode character of "the finger"?


? ?
peek...
Feb 02 2004
parent "Serge K" <skarebo programmer.net> writes:
	charset="utf-8"
Content-Transfer-Encoding: quoted-printable

grr... encodings...
=E2=9C=8C =E2=98=A0

270Ch, 2620h
Feb 02 2004
prev sibling next sibling parent reply Juan C <Juan_member pathlink.com> writes:
I prefer auto-declared "result" variable like Eiffel or Delphi.


<snobbery>
It seems to me that "scripting" languages do this, and "programming" languages
require declaration of variables. D is a programming language, and I am a
programmer, not a script writer.

Although some scripting languages are now quite powerful (Perl and Java), and

a distinction and times when a scripting language just won't cut it.
</snobbery>
Jan 31 2004
parent reply Roel Mathys <roel.mathys yucom.be> writes:
Juan C wrote:

I prefer auto-declared "result" variable like Eiffel or Delphi.


 
 
 <snobbery>
 It seems to me that "scripting" languages do this, and "programming" languages
 require declaration of variables. D is a programming language, and I am a
 programmer, not a script writer.
 
 Although some scripting languages are now quite powerful (Perl and Java), and

still
 a distinction and times when a scripting language just won't cut it.
 </snobbery>
 
 


Delphi is a compiled language and not a "scripting" language, don't know 
whether this is standard pascal though.
btw: its compile times are fast, really, really fast.

I don't like Java, but it's rarely spoken about as a scripting language. 
And if I'm not mistaken efforts are underway to generate a native binary 
from Java source code.

My scripting language of choice would be Python :-)

bye,
roel
Feb 01 2004
parent reply Stephan Wienczny <wienczny web.de> writes:
Roel Mathys wrote:

 
 Delphi is a compiled language and not a "scripting" language, don't know 
 whether this is standard pascal though.
 btw: its compile times are fast, really, really fast.


Delphi uses a language called Object Pascal.
It uses modules and some kind of precompilation.

Stephan
Feb 01 2004
parent reply Roel Mathys <roel.mathys yucom.be> writes:
Stephan Wienczny wrote:


 Roel Mathys wrote:
 

 Delphi is a compiled language and not a "scripting" language, don't 
 know whether this is standard pascal though.
 btw: its compile times are fast, really, really fast.


 
 Delphi uses a language called Object Pascal.
 It uses modules and some kind of precompilation.
 
 Stephan
 


you mean each unit is precompiled into an object file and only 
recompiled whenever the source in it has changed?
This speeds up the *generation* process, just like VC++ with precompiled 
headers and other languages that use some kind of modules :-)
This way compilation only happens for a few files, but linking still 
needs to be done.

or do you mean don't
"don"t know whether this is standard pascal though"
=> I knew delphi = pascal with (borland property) object extension

In the meantime I looked it up in one of my old (and I mean old) Pascal 
manual, and the "result" shortcut was certainly in it.

bye,
roel
Feb 01 2004
parent Georg Wrede <Georg_member pathlink.com> writes:
In article <bvingb$27ve$1 digitaldaemon.com>, Roel Mathys says...

Stephan Wienczny wrote:


 Roel Mathys wrote:
 

 Delphi is a compiled language and not a "scripting" language, don't 
 know whether this is standard pascal though.
 btw: its compile times are fast, really, really fast.






Ever since Turbo Pascal 1.0 (23 years ago) they've been
very proud of having the fastest Pascal compilers, and 
because of the language itself, also faster than any C
compiler.

Object Pascal, as they call the language, is the de facto 
standard today. (Even if Richard Stallman &co try 
to belittle this factoid.)


 Delphi uses a language called Object Pascal.
 It uses modules and some kind of precompilation.




The Delphi compiler is split in two stages, a "front" and a
"back". The "back" is the same for Delphi and Borland C++.
(I guess Walter does the same with D and his C++ ??)


In the meantime I looked it up in one of my old (and I mean old) Pascal 
manual, and the "result" shortcut was certainly in it.


I've always been under the impression that this was one of
the original thoughts when N. Wirth published the language.
One of the main ideas was to cut through unnecessary writing,
where possible. The language was to be smooth, consistent,
easy to learn, yet powerful enough for serious programming.

Pascal is actually older than C. It was developed during the
late 1960's.

In my bookshelf I found "The Pascal Language" published by Helsinki
University of Technology - Laboratory of Information Processing
Science, 1980. The "result shortcut" is presented as an integral
part of Pascal. At the end is an appendix about differences
between some implementations of the day, and the result shortcut
is not among them.
Feb 01 2004
prev sibling next sibling parent reply Andy Friesen <andy ikagames.com> writes:
Y.Tomino wrote:


 
 And return is one of jump statements.
 Although I don't dislike a jump statement, I don't want to use it where jump
 is unnecessary.
 Even if result introduced, the compatibility will be kept if return will set
 result before exit.
 
 Thanks.
 YT
 


I think everybody can agree that it's just sugar, which is fine, really. 
  Everything above machine code is technically sugar, it's just a matter 
of how much work that sugar does for you.

Is there any case where adding this to the language would save more than 
two lines of code?  Two very simple, obvious lines of code, no less.  I 
don't think so.  So, to be blunt, why bother?

  -- andy
Feb 01 2004
next sibling parent reply "Y.Tomino" <demoonlit inter7.jp> writes:
Everyone will call "return" as syntax sugar, supposing "result" was
implemented previously.
I'm sorry.

YT
Feb 01 2004
parent Andy Friesen <andy ikagames.com> writes:
Y.Tomino wrote:

 Everyone will call "return" as syntax sugar, supposing "result" was
 implemented previously.


Right.  Either can be easily simulated with the other.


 I'm sorry.
 


Don't be.  It's better to have the ideas out in the open than never even 
considered.

  -- andy
Feb 01 2004
prev sibling next sibling parent Juan C <Juan_member pathlink.com> writes:
In article <bvjken$k2t$1 digitaldaemon.com>, Andy Friesen says...

Y.Tomino wrote:


 
 And return is one of jump statements.
 Although I don't dislike a jump statement, I don't want to use it where jump
 is unnecessary.
 Even if result introduced, the compatibility will be kept if return will set
 result before exit.
 
 Thanks.
 YT
 


I think everybody can agree that it's just sugar, which is fine, really. 
  Everything above machine code is technically sugar, it's just a matter 
of how much work that sugar does for you.

Is there any case where adding this to the language would save more than 
two lines of code?  Two very simple, obvious lines of code, no less.  I 
don't think so.  So, to be blunt, why bother?

  -- andy


And so it should be possible for an IDE to not only add the braces, but the
declaration and return statements as well. Howsabout that?
Feb 02 2004
prev sibling parent reply matthias becker <matthias_member pathlink.com> writes:
Is there any case where adding this to the language would save more than 
two lines of code?  Two very simple, obvious lines of code, no less.  I 
don't think so.  So, to be blunt, why bother?


Hey it's two lines and two lines are two lines. As my methods are commonly very
short two lines do matter.
Feb 05 2004
parent J C Calvarese <jcc7 cox.net> writes:
matthias becker wrote:

Is there any case where adding this to the language would save more than 
two lines of code?  Two very simple, obvious lines of code, no less.  I 
don't think so.  So, to be blunt, why bother?


 
 
 Hey it's two lines and two lines are two lines. As my methods are commonly very
 short two lines do matter.


I've been in the habit of writing long functions. I'm trying process 
data using shorter functions these days, but two lines still doesn't 
seem like it could be a huge savings. Can you give an example of a 
typical short function that you'd write (that would benefit from using 
result)? You could do a before and after, maybe?

I've written short functions like this ...

char[] toString()
{
   if(mvalue)
     return "True";
   else
     return "False";
}

or

bit value(){ return mvalue; }

but they wouldn't even benefit from "result".

Also, I'm not convinced we could be saving two lines.  It seems to me 
we're only really saving one line.  I think a real example (no ...'s) 
could help me understand the actual benefit of this proposal.

My two cents...

-- 
Justin
http://jcc_7.tripod.com/d/
Feb 05 2004
prev sibling parent "C" <dont respond.com> writes:
I always liked this from delphi , but as has been discussed prolly not for
D.  Good ideas though keep em coming.

C

"Y.Tomino" <demoonlit inter7.jp> wrote in message
news:bvhb5v$2uc4$1 digitaldaemon.com...

 Many functions are written as follows:

 int func(...)
 {
   int result;
   ...
   result = ...
   ...
   return result;
 }

 I feel it's needless bother.
 I prefer auto-declared "result" variable like Eiffel or Delphi.

 int func(...)
 {
   ...
   result = ...
   ...
 }

 It simplifies code.

 And return is one of jump statements.
 Although I don't dislike a jump statement, I don't want to use it where


jump

 is unnecessary.
 Even if result introduced, the compatibility will be kept if return will


set

 result before exit.

 Thanks.
 YT
Feb 01 2004