I suspect Walter meant "work to do by hand"?


"Anders F Björklund" <afb@algonet.se> wrote in message news:dsb971$68u$1@digitaldaemon.com...
> Walter Bright wrote:
>
>>>From the Qt# home page, they say there are 476 classes converted...
>>
>> That's just one heck of a lot of work to do.
>
> I don't how many classes there are in SWT, and maybe not all of them are needed for D (some might be Java specific), but if effectively triples* if you now plan on porting all of them over to D (as opposed to just porting C++ wrappers)
>
> So I would think that DWT will end up having many *more* ?
>
> --anders
>
> * assuming you want to support three platforms
>   (Windows, GTK, Mac OS X). Multiply further,
>   for supporting any other wanted platforms...

pragma schrieb am 2006-02-07:
> In article <dsafo1$2c19$1@digitaldaemon.com>, Don Clugston says...
>>
>>rant
>>Personally, I think the default LGPL license is the most ridiculous one
>>I've seen, "Arse licence" included. The dynamic linking requirement
>>doesn't benefit anyone. It's annoying for commercial use, but it doesn't
>>benefit open source at all. GPL makes a lot of sense (in the right
>>context, but definitely not for D standard libraries), but LGPL is just
>>nuts.
>>/rant
>
> You're not alone in thinking this.  I myself find all GPL variants rather dogmatic, and practically a form of "PM-repellant" around the office.  I always felt the tone of the license to be more on the side of foisting a particular agenda on developers rather than promoting "freedom".  The current flap surrounding GPLv3, and the proliferation of other more flexible licenses, has only solidified this belief of mine.
>
> - Eric Anderton at yahoo

This might apply if the software is bought.

However if you buy services - classic examples are webhosting and feature adding - (L)GPL can be convinent.

Thomas

On Tue, 7 Feb 2006 03:05:14 -0800, Walter Bright wrote:

> "Derek Parnell" <derek@psych.ward> wrote in message
>> * Why don't you want others to recompile Phobos using a heightened measure of caution (ie "-w" switch)
> 
> In the previous long, and heated, discussions on this issue, I argued passionately that the -w style code reduces readability, maintainability, and reliability. I don't think there's anything to be added to that.

But I do <g>.  I have a different opinion from your opinion. My opinion is that in those cases where the compiler makes an assumption of my intentions, I can reduce potentially incorrect assumptions if I annotate my code with explicit instructions to the compiler about my actual intentions.

So when I code

   int a;
   long b;
   . . .
   a = b;

Instead of the compiler assuming I meant ...

   a = cast(int)b;

with the "-w" switch I could be alerted to this compiler assumption and recode it to be exactly explicit. Maybe I would code it as ...

   if (b > a.max )
      throw (SomeERROR);
   a = cast(int)b;

or maybe ...

   long a;
   long b;
   . . .
   a = b;

or maybe ..., well you get the picture. There could be any number of alternatives that are *not* what the compiler assumed on my behalf.

> This current thread only reinforces my opinion that the mere existence of warnings is a mistake in any language, as it causes unresolvable debates about "is this code right or wrong?"

I am *NOT* talking about right or wrong, or even "errors". I just simply want to know where it is that the compiler is assuming what my intentions might have been so that I can recode it to be explicit. Nothing whatsoever to do with right or wrong, legal or illegal, etc...

If you are concerned about the term "warning", then for Bob's sake change the terminology. I don't care that much about what term you are using for this concept.

-- 
Derek
(skype: derek.j.parnell)
Melbourne, Australia
"Down with mediocracy!"
8/02/2006 12:06:35 PM

Kris wrote:

>>>>From the Qt# home page, they say there are 476 classes converted...
>>>
>>>That's just one heck of a lot of work to do.
[...]

>>So I would think that DWT will end up having many *more* 

> I suspect Walter meant "work to do by hand"?

He might have...

Anyway, there's a pretty big difference between wrappers such
as "QtD" where most things can either be autogenerated in the
same manner as Qt#/QtC or converted over to D from those files,
and where most of the actual OS interface code is hidden away
in the C++ libraries - when compared to something like SWT/DWT,
where as little as possible is "extern" as it has the code too ?

But if the translation can be automated, then maybe it's not
that much harder translating the SWT source code ? I don't have
much insight as to have the Win32 port was accomplished, really.
And I only know of the posting to the DWT forum, for the others:
("porting" : http://dsource.org/forums/viewtopic.php?t=1169)
And that didn't really sound too good for the GTK/Mac ports... ?


Guess it all comes down to if you want a GUI lib *written* in D,
or if you just want a GUI *usable* from D code (i.e. external/C++)

(I'm assuming that DWT won't use any external libs, except for OS)

--anders

On Tue, 7 Feb 2006 13:51:05 -0800, Walter Bright wrote:

> "Kramer" <Kramer_member@pathlink.com> wrote in message news:dsalta$2ivu$1@digitaldaemon.com...
>> So why not instead treat the compiler -w option as a LINT like tool.  I
>> know
>> it's *not* LINT, but maybe it's just a terminology thing that's hanging
>> things
>> up.  Treat -w like LINT and allow the compiler to compile through the
>> 'warnings'.  Maybe instead of 'warnings', they should be 'code that you
>> are
>> highly advised to look at, because the compiler doesn't agree with you'.
> 
> The trouble with that is it sounds good on paper, but in practice those warnings scroll up and off the screen, and are missed.

What sort of console software do you use? I use Windows XP cmd.exe with the buffer set to 9999 lines, and I've never had warnings etc... scroll off the screen. And if one does, you can redirect to a file and examine them at leisure.

-- 
Derek
(skype: derek.j.parnell)
Melbourne, Australia
"Down with mediocracy!"
8/02/2006 12:30:54 PM

Anders F Björklund wrote:
> 
> But if the translation can be automated, then maybe it's not
> that much harder translating the SWT source code ? I don't have
> much insight as to have the Win32 port was accomplished, really.
> And I only know of the posting to the DWT forum, for the others:
> ("porting" : http://dsource.org/forums/viewtopic.php?t=1169)
> And that didn't really sound too good for the GTK/Mac ports... ?

I looked into automated C porting recently and it seems achievable, but you basically need the tool to contain a modified macro preprocessor.  I think what's needed is to convert all #if blocks to static if blocks and leave the conditions in place, but expand all macro functions.  I've been rooting around for a simple parser to adapt for this purpose to ease porting POSIX headers.

> Guess it all comes down to if you want a GUI lib *written* in D,
> or if you just want a GUI *usable* from D code (i.e. external/C++)

Written in D is obviously ideal.

> (I'm assuming that DWT won't use any external libs, except for OS)

Out of curiosity, does SWT work on any non-standard architectures? Could I write an app using SWT and run it on a PocketPC?


Sean

"Sean Kelly" <sean@f4.ca> wrote in..
> Out of curiosity, does SWT work on any non-standard architectures? Could I write an app using SWT and run it on a PocketPC?

From the prior links: http://www.eclipse.org/articles/Article-small-cup-of-swt/pocket-PC.html

The Win32 version is a superset of the PocketPC one. However, you'd need a GDC targeting ARM/XScale or H8S.

Walter Bright wrote:
> Can we move this discussion over to digitalmars.D.dwt?
> 

Have you thought about moving to a PHPBB type forum? It would be much more accessible to newcomers. Just a thought.

"nick" <nick.atamas@gmail.com> wrote in message news:dsbr0b$l45$1@digitaldaemon.com...
> Walter Bright wrote:
>> Can we move this discussion over to digitalmars.D.dwt?
>>
>
> Have you thought about moving to a PHPBB type forum? It would be much more accessible to newcomers. Just a thought.

Yes, there have been at least a couple of discussions about it.

"Derek Parnell" <derek@psych.ward> wrote in message news:adnet1wo97n0$.fruquqppxjn$.dlg@40tude.net...
> On Tue, 7 Feb 2006 13:51:05 -0800, Walter Bright wrote:
>> The trouble with that is it sounds good on paper, but in practice those warnings scroll up and off the screen, and are missed.
> What sort of console software do you use? I use Windows XP cmd.exe with
> the
> buffer set to 9999 lines, and I've never had warnings etc... scroll off
> the
> screen. And if one does, you can redirect to a file and examine them at
> leisure.

Oh, I know how to do both of those things. The trouble is, when you're doing a long compile, unless you see it on the screen you don't think to go look at a log file or whatever. Having the build not complete means you can't overlook it.