www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - D vs C Char

reply Trevor Parscal <trevorparscal hotmail.com> writes:
OK, so I am still having trouble with this FreeType library, and ... The 
  people from the FreeType newsgroup told me..


- do not use the D 'char' for C 'char'. That's because
   it's default initializer is 0xFF instead of 0, and this
   will cause all kinds of subtle problems when used
   naively in D code.

So what do i use instead? byte? Isn't the intializer the same?

-- 
Thanks,
Trevor Parscal
www.trevorparscal.com
trevorparscal hotmail.com
Jun 29 2005
next sibling parent "Unknown W. Brackets" <unknown simplemachines.org> writes:
I would use ubyte...

-[Unknown]



 OK, so I am still having trouble with this FreeType library, and ... The 
  people from the FreeType newsgroup told me..
 
 
 - do not use the D 'char' for C 'char'. That's because
   it's default initializer is 0xFF instead of 0, and this
   will cause all kinds of subtle problems when used
   naively in D code.
 
 So what do i use instead? byte? Isn't the intializer the same?
Jun 29 2005
prev sibling next sibling parent reply Brad Beveridge <brad somewhere.net> writes:
Trevor Parscal wrote:

 OK, so I am still having trouble with this FreeType library, and ... The 
  people from the FreeType newsgroup told me..
 
 
 - do not use the D 'char' for C 'char'. That's because
   it's default initializer is 0xFF instead of 0, and this
   will cause all kinds of subtle problems when used
   naively in D code.
 
 So what do i use instead? byte? Isn't the intializer the same?
 


That smells like a bogus answer to me.  Nobody in their right mind 
should rely on a C compiler to initialise variables.
The only difference between C's char and D's char is that D explicitly 
defines char to be unsigned, a C compiler may choose to define it as 
signed or unsigned, depending on architecture/whim.

If there is truely a difference between the default init value, then you 
can explicitly set the variable.

char c = 0;

BTW, have you got a simple working C example of Freetype that works? 
Can you convert that to D?  I'm wondering if you're having troubles 
because of misusing the Freetype API rather than the C/D interface.

Brad
Jun 29 2005
parent reply Trevor Parscal <trevorparscal hotmail.com> writes:
Brad Beveridge wrote:

 Trevor Parscal wrote:
 

 OK, so I am still having trouble with this FreeType library, and ... 
 The  people from the FreeType newsgroup told me..


 - do not use the D 'char' for C 'char'. That's because
   it's default initializer is 0xFF instead of 0, and this
   will cause all kinds of subtle problems when used
   naively in D code.

 So what do i use instead? byte? Isn't the intializer the same?


 That smells like a bogus answer to me.  Nobody in their right mind 
 should rely on a C compiler to initialise variables.
 The only difference between C's char and D's char is that D explicitly 
 defines char to be unsigned, a C compiler may choose to define it as 
 signed or unsigned, depending on architecture/whim.
 
 If there is truely a difference between the default init value, then you 
 can explicitly set the variable.
 
 char c = 0;
 
 BTW, have you got a simple working C example of Freetype that works? Can 
 you convert that to D?  I'm wondering if you're having troubles because 
 of misusing the Freetype API rather than the C/D interface.
 
 Brad


Thats a good question.. I suppose I should try and do a test with C.. 
But, its allot of configuration crap just to get FreeType to compile 
with C at all.. so.. I have been avoiding dealing with that.

They say on the FreeType page, "FreeType will compile on almost any 
compiler on most platforms without even a single warning".. But, my 
experience is that it's EXTRMELY difficult to get it to compile at all, 
and there are so many warnings its hard to see where the errors are.

Anyhoo, I cant even begin to explain how much I wish this worked, and 
would give my semicolin key to the man who helps me do so.

Thanks,
Trevor Parscal

-- 
Thanks,
Trevor Parscal
www.trevorparscal.com
trevorparscal hotmail.com
Jun 29 2005
parent Brad Beveridge <brad somewhere.net> writes:
Trevor Parscal wrote:
<snip>

 
 Thats a good question.. I suppose I should try and do a test with C.. 
 But, its allot of configuration crap just to get FreeType to compile 
 with C at all.. so.. I have been avoiding dealing with that.
 
 They say on the FreeType page, "FreeType will compile on almost any 
 compiler on most platforms without even a single warning".. But, my 
 experience is that it's EXTRMELY difficult to get it to compile at all, 
 and there are so many warnings its hard to see where the errors are.
 
 Anyhoo, I cant even begin to explain how much I wish this worked, and 
 would give my semicolin key to the man who helps me do so.
 
 Thanks,
 Trevor Parscal
 


The documentation has some snippet code examples that look like they'd 
go together without much trouble.  I did notice one interesting thing, 
the docs say you should
#include <ftbuild2.h>
#include FT_FREETYPE_H

which looks a little weird to me, I think they are doing some non 
standard things, and they mention that they are doing magic trickery.

I think that my approach to this would be
1) Write a C program that gets free type working
2) Convert that program to something more like a library, so that it 
proxies the free type calls that you are actually interested in.
3) Interface your D code to this proxy library/object file.
4) Start directly interfacing to Freetype from D where appropriate.

Unless I understood what was going on in the headers really well, I'd 
try and take the above approach.

Cheers
Brad
Jun 29 2005
prev sibling next sibling parent reply =?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb algonet.se> writes:
Trevor Parscal wrote:


 - do not use the D 'char' for C 'char'. That's because
   it's default initializer is 0xFF instead of 0, and this
   will cause all kinds of subtle problems when used
   naively in D code.
 
 So what do i use instead? byte? Isn't the intializer the same?


No, as the various D integer initializers are still sane... :-)
It's just the code units and floating points that's "wild".

(As per the table at http://www.digitalmars.com/d/type.html)


It all depends on what they are using "char" for, really ?
If it's ASCII, then _either_ "char" or "ubyte" will do fine.
If they are using char for 8-bit encodings, then it's ubyte.
And if they are using char for UTF-8 (unlikely), then: char.

And for "char*", it doesn't matter much at all really (since
you get a pointer), so most people just stick with the C def.
Even if it's technically wrong in D, since "char*" means UTF-8.
(and not "some random encoding", like its the old C meaning is)


That being said,
I haven't had a problem translating C's "char*" to D's "char*",
in any of the import modules for D that I have done thus far...

I'll feed the freetype2 headers to my C header->D module script,
and see what it comes up with. Minus the weirdo macros, that is.

But don't hold your breath. ;-)
--anders
Jun 29 2005
parent reply Trevor Parscal <trevorparscal hotmail.com> writes:
Anders F Björklund wrote:


 I'll feed the freetype2 headers to my C header->D module script,
 and see what it comes up with. Minus the weirdo macros, that is.
 
 But don't hold your breath. ;-)
 --anders


I would be VERY interested in seeing what you come up with.. Send me a 
copy if you can once it is under control, or when you give up...

-- 
Thanks,
Trevor Parscal
www.trevorparscal.com
trevorparscal hotmail.com
Jun 29 2005
next sibling parent reply =?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb algonet.se> writes:
Trevor Parscal wrote:


 I would be VERY interested in seeing what you come up with.. Send me a 
 copy if you can once it is under control, or when you give up...


They are using macros for *everything*, including names of headers -
so it needs to be preprocessed after conversion... <mutter, mutter>

(I'm not preprocessing before conversion, since the #defines usually
holds a lot of usable information that can be converted into D things)

Wasn't very hard to compile, though ? "./configure && make". Yay, UNIX.

--anders
Jun 29 2005
parent reply Trevor Parscal <trevorparscal hotmail.com> writes:
Anders F Björklund wrote:

 Trevor Parscal wrote:
 

 I would be VERY interested in seeing what you come up with.. Send me a 
 copy if you can once it is under control, or when you give up...


 
 
 They are using macros for *everything*, including names of headers -
 so it needs to be preprocessed after conversion... <mutter, mutter>
 
 (I'm not preprocessing before conversion, since the #defines usually
 holds a lot of usable information that can be converted into D things)
 
 Wasn't very hard to compile, though ? "./configure && make". Yay, UNIX.
 
 --anders


yes.. Yay UNIX, BOOO Windows...

-- 
Thanks,
Trevor Parscal
www.trevorparscal.com
trevorparscal hotmail.com
Jun 29 2005
parent reply Mike Parker <aldacron71 yahoo.com> writes:
Trevor Parscal wrote:

 
 
 yes.. Yay UNIX, BOOO Windows...
 


Install Cygwin or MingW + MSYS on Windows and you can do the same thing: 
./configure
make
Jun 29 2005
parent reply brad beveridge <brad nowhere.com> writes:
Mike Parker wrote:

 Trevor Parscal wrote:
 

 yes.. Yay UNIX, BOOO Windows...


 
 Install Cygwin or MingW + MSYS on Windows and you can do the same thing: 
 ./configure
 make


True, but in the past I have found that cygwin is a poor substitute for 
the real thing :)

Brad
Jun 29 2005
parent Trevor Parscal <trevorparscal hotmail.com> writes:
brad beveridge wrote:

 Mike Parker wrote:
 

 Trevor Parscal wrote:


 yes.. Yay UNIX, BOOO Windows...


 Install Cygwin or MingW + MSYS on Windows and you can do the same 
 thing: ./configure
 make


 
 True, but in the past I have found that cygwin is a poor substitute for 
 the real thing :)
 
 Brad


Yes, MSSYS and ctgwin have both been less than satisfying..

besides, compiling on linux / unix isn't what I needed to do with 
FreeType... I needed to compile it on Windows, for windows... thus..

BOOO Windows... :)

-- 
Thanks,
Trevor Parscal
www.trevorparscal.com
trevorparscal hotmail.com
Jun 29 2005
prev sibling parent =?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb algonet.se> writes:
Trevor Parscal wrote:


 I'll feed the freetype2 headers to my C header->D module script,
 and see what it comes up with. Minus the weirdo macros, that is.

 But don't hold your breath. ;-)


 
 I would be VERY interested in seeing what you come up with.. Send me a 
 copy if you can once it is under control, or when you give up...


No go, sorry. There's just too many macros used (worse than SDL!)
So I think you are better off with your hand-translated version ?
(last time I heard, you were still trying to get it to link with
a test C program or something, so that would probably be step one)

I have a similar problem here on Mac OS X, with the "Carbon" headers
that use some C features which makes them hard to port over to D :-(
They work just fine from C, C++, and Objective-C too; but require some
hand-holding to work with D. I've done selected functions, but not all.


Unfortunately, this makes D much less useful - when it can't use C libs.
It's of course doable to write the import modules, just a *lot* of work?
Another problem in this particular case is that they are copyrighted, so
even if I do get them ported I'm not sure I'm allowed to redistribute...

But at least this is one of the "known issues" with the D language.
" These are some of the major bugs:
   Need to write a tool to convert C .h files into D imports. "
(From http://www.digitalmars.com/d/dcompiler.html#general)

--anders
Jun 30 2005
prev sibling parent Bruno Medeiros <daiphoenix NOSPAMlycos.com> writes:
Trevor Parscal wrote:

 OK, so I am still having trouble with this FreeType library, and ... The 
  people from the FreeType newsgroup told me..
 
 
 - do not use the D 'char' for C 'char'. That's because
   it's default initializer is 0xFF instead of 0, and this
   will cause all kinds of subtle problems when used
   naively in D code.
 
 So what do i use instead? byte? Isn't the intializer the same?
 


May I ask you why are you using the FreeType library? And also why are 
you compiling it, is that necessary to use it?

-- 
Bruno Medeiros
Computer Science/Engineering student
Jul 03 2005