digitalmars.D - version - feature suggestions

=?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb algonet.se> Oct 18 2004

"Ivan Senji" <ivan.senji public.srce.hr> Oct 18 2004

=?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb algonet.se> Oct 18 2004

"Walter" <newshound digitalmars.com> Oct 19 2004

=?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb algonet.se> Oct 20 2004

Sean Kelly <sean f4.ca> Oct 19 2004
"Walter" <newshound digitalmars.com> Oct 19 2004

=?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb algonet.se> Oct 20 2004

Regan Heath <regan netwin.co.nz> Oct 20 2004

J C Calvarese <jcc7 cox.net> Oct 21 2004

Regan Heath <regan netwin.co.nz> Oct 21 2004

Burton Radons <burton-radons smocky.com> Oct 21 2004

=?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb algonet.se> writes:

Two features seems to be missing from versions:

1) reversing the condition

This C maps rather easily into D:

 #ifdef SOMETHING
 // do stuff
 #endif


as:

 version (SOMETHING) {
 // do stuff
 }


This does not:

 #ifndef SOMETHING
 // do stuff
 #endif


The current D workaround:

 version (SOMETHING)
   version = SOMETHING;
 else
   version = not_SOMETHING;
 
 version (not_SOMETHING) {
 // do stuff
 }


That quickly gets boring to write...

* Suggestion: version (!SOMETHING)


2) numbered versions

This code is kinda hard to "translate" :

 #define VERSION 2
 
 #if (VERSION > 2)
 // do stuff
 #endif


Current workaround in D:

 version = VERSION_1;
 version = VERSION_2;
 
 version (VERSION_2) {
 // do stuff 
 }


Again, it gets boring around 10 or so.

* Suggestion: version (VERSION > 2)


What do you think ? (am I missing something?)

--anders


PS. Another little ugliness is the casing:
     "Windows" versus "linux" versus "darwin"

     Perhaps we should ask the compilers to
     define *both* lowercase and CamelCase ?

Oct 18 2004

"Ivan Senji" <ivan.senji public.srce.hr> writes:

"Anders F Bj�rklund" <afb algonet.se> wrote in message
news:cl0s4t$hs1$1 digitaldaemon.com...
 Two features seems to be missing from versions:

 1) reversing the condition

 This C maps rather easily into D:

 #ifdef SOMETHING
 // do stuff
 #endif


 as:

 version (SOMETHING) {
 // do stuff
 }


 This does not:

 #ifndef SOMETHING
 // do stuff
 #endif


 The current D workaround:

 version (SOMETHING)
   version = SOMETHING;
 else
   version = not_SOMETHING;

 version (not_SOMETHING) {
 // do stuff
 }




Can't you write that as
version(SOMETHING){}
else
{
    // do stuff
}


 ...

Oct 18 2004

=?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb algonet.se> writes:

Ivan Senji wrote:

The current D workaround:

version (SOMETHING)
  version = SOMETHING;
else
  version = not_SOMETHING;

version (not_SOMETHING) {
// do stuff
}




 Can't you write that as
 version(SOMETHING){}
 else
 {
     // do stuff
 }


Hehe, yup - you are perfectly right there ! :-)

 #ifndef SOMETHING
 // do stuff
 #endif


becomes

 version (SOMETHING) {} else {
 // do stuff
 }


A little obscure, but better than my workaround...


The only problem is that I now have the following:

 #ifndef SOMETHING
 // do stuff
 #else
 // do something else
 #endif


I'd prefer *not* to swap them, since it's ported C.
The previous workaround still holds here, though:

 version (not_SOMETHING) {
 // do stuff
 } else {
 // do something else
 }


I can use the short workaround for most, and long for rest.

--anders

PS. It was for glext.h

Oct 18 2004

"Walter" <newshound digitalmars.com> writes:

"Anders F Bj�rklund" <afb algonet.se> wrote in message
news:cl17lu$ujq$1 digitaldaemon.com...
 Can't you write that as
 version(SOMETHING){}
 else
 {
     // do stuff
 }


 Hehe, yup - you are perfectly right there ! :-)


<g>. Actually, the idea here is that versions should represent something
positive, like DO_THIS_FEATURE rather than "DO_NOT_DO_THIS_FEATURE".

Oct 19 2004

=?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb algonet.se> writes:

Walter wrote:

 <g>. Actually, the idea here is that versions should represent something
 positive, like DO_THIS_FEATURE rather than "DO_NOT_DO_THIS_FEATURE".


They were being used as:

if (SANE) {} else {
// activate ugly workaround
}

So it was kinda positive ? :)

--anders

Oct 20 2004

Sean Kelly <sean f4.ca> writes:

Anders F Bj�rklund wrote:
 Two features seems to be missing from versions:
 
 1) reversing the condition
 
 This C maps rather easily into D:
 
 #ifdef SOMETHING
 // do stuff
 #endif


 
 as:
 
 version (SOMETHING) {
 // do stuff
 }


 
 This does not:
 
 #ifndef SOMETHING
 // do stuff
 #endif




It would be nice to have full expression logic in version and debug 
statements, assuming we don't already.


Sean

Oct 19 2004

"Walter" <newshound digitalmars.com> writes:

"Anders F Bj�rklund" <afb algonet.se> wrote in message
news:cl0s4t$hs1$1 digitaldaemon.com...
 PS. Another little ugliness is the casing:
      "Windows" versus "linux" versus "darwin"


These casings were picked to match the casings used by the C compilers on
the corresponding systems.

Oct 19 2004

=?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb algonet.se> writes:

Walter wrote:

PS. Another little ugliness is the casing:
     "Windows" versus "linux" versus "darwin"


 These casings were picked to match the casings used 
 by the C compilers on the corresponding systems.


Couldn't *both* versions be defined ? As in:

windows,Windows,linux,Linux,darwin,Darwin, etc.


Since both Win32 and Windows are being used now ?

Also suggested a "MacOSX" to David Friedman (gdc)

--anders

PS. It's mostly an aesthetic issue,
     although loader.d was broken...
     (as it uses version(Linux) now)

Oct 20 2004

Regan Heath <regan netwin.co.nz> writes:

On Wed, 20 Oct 2004 14:47:53 +0200, Anders F Bj�rklund <afb algonet.se> 
wrote:
 Walter wrote:

 PS. Another little ugliness is the casing:
     "Windows" versus "linux" versus "darwin"


 These casings were picked to match the casings used by the C compilers 
 on the corresponding systems.


 Couldn't *both* versions be defined ? As in:

 windows,Windows,linux,Linux,darwin,Darwin, etc.


 Since both Win32 and Windows are being used now ?

 Also suggested a "MacOSX" to David Friedman (gdc)

 --anders

 PS. It's mostly an aesthetic issue,
      although loader.d was broken...
      (as it uses version(Linux) now)


I'd prefer if it was simply case insensitive.

I know D is case sensitive, and that we like to have consistency, but, I 
think this is another instance where breaking from the trend is the 
RightThing(TM).

Regan

-- 
Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/

Oct 20 2004

J C Calvarese <jcc7 cox.net> writes:

Regan Heath wrote:
 On Wed, 20 Oct 2004 14:47:53 +0200, Anders F Bj�rklund <afb algonet.se> 


 PS. It's mostly an aesthetic issue,
      although loader.d was broken...
      (as it uses version(Linux) now)


 
 I'd prefer if it was simply case insensitive.
 
 I know D is case sensitive, and that we like to have consistency, but, I 
 think this is another instance where breaking from the trend is the 
 RightThing(TM).
 
 Regan


I don't want the version identifiers to be case insensitive, but there's 
no good reason why the the predefined versions shouldn't be consistent.


Mixed case (I prefer this)...
version(Windows)
version(Linux)
version(MacOSX)
version(Darwin)


or all lowercase...
version(windows)
version(linux)
version(macosx)


Some of each hurts my eyes...
version(Windows)
version(linux)

You might as well write, "Open the Door to let roger in."

-- 
Justin (a/k/a jcc7)
http://jcc_7.tripod.com/d/

Oct 21 2004

Regan Heath <regan netwin.co.nz> writes:

On Thu, 21 Oct 2004 22:34:58 -0500, J C Calvarese <jcc7 cox.net> wrote:
 Regan Heath wrote:
 On Wed, 20 Oct 2004 14:47:53 +0200, Anders F Bj�rklund <afb algonet.se>


 PS. It's mostly an aesthetic issue,
      although loader.d was broken...
      (as it uses version(Linux) now)



 I'd prefer if it was simply case insensitive.

 I know D is case sensitive, and that we like to have consistency, but, 
 I think this is another instance where breaking from the trend is the 
 RightThing(TM).

 Regan


 I don't want the version identifiers to be case insensitive,


Why not?

 but there's no good reason why the the predefined versions shouldn't be 
 consistent.


At least if they're consistent we don't have to guess what they are.


My reasoning for making them case insensitive is that it becomes less 
possible for a typo to cause trouble.

i.e.

version (windows)
version (WINDOWS)
version (Windows)
version (WiNdOwS)

will all work. No guessing, just spell it write (joke intentional) and it 
will work as intended.

Regan.

-- 
Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/

Oct 21 2004

Burton Radons <burton-radons smocky.com> writes:

Walter wrote:
 "Anders F Bj�rklund" <afb algonet.se> wrote in message
 news:cl0s4t$hs1$1 digitaldaemon.com...
 
PS. Another little ugliness is the casing:
     "Windows" versus "linux" versus "darwin"


 
 These casings were picked to match the casings used by the C compilers on
 the corresponding systems.


All that that means is that the C method was an ungodly mess that should 
be sorted out by a sensible and consistent New Way:

     Compiler.DigitalMars
     Compiler.GDC
     Compiler.DNET
     Processor.X86
     Processor.AMD64
     Processor.Endian.Big
     Processor.Endian.Little
     OS.Windows
     OS.Linux
     Asm.X86
     None

There should also be versions describing such details as what kind of 
calling convention is used by the compiler for D linking to reduce 
potential for assumptions (false assumptions are the cause of all 
accidents.)  As it is, there is no way to prove that an asm statement is 
safe; therefore, they cannot be put in robust libraries.

Oct 21 2004

D Programming

C/C++ Programming

Other

digitalmars.D - version - feature suggestions