OpenEuphoria

struct rect """ 

typedef struct _RECT { 
  LONG left; 
  LONG top; 
  LONG right; 
  LONG bottom; 
} RECT, *PRECT; 
""" 

end struct 
rect q = new({10,10,350,200}), 
     r = new() 
r.left = 10     --\   equivalent to using the 
r.top = 10      -- }   {10,10,350,200} above 
r.right = 350   -- }  and/or m = allocate(16) 
r.bottom = 200  --/    & poke4(m+0/4/8/12,..). 
 
-- then eg c_proc(xFillRect,{hDC,struct_mem(r|q)|m,hBrush}) 

D:\testpool\ray_test\struct_test.ex:10 
type check failure, q is {"struct","rect",2,10910404.0} 
 
Global & Local Variables 
 
--> see D:\testpool\ray_test\ex.err 
Press Enter... 

2. Re: Phix: c-struct

• Posted by petelomax 2 weeks ago
• 213 views

Oh dear. That works fine in 0.8.2 but is broken in 1.0.1 and everything after...
Obviously no-one, including myself, has used c-structs in the last five years, so I'm inclined to just drop 'em.
Anyway, a c-struct was only ever meant to be a thin wrapper to the following, which still works fine.

include cffi.e 
constant idRECT = define_struct("""  

typedef struct _RECT {  
  LONG left;  
  LONG top;  
  LONG right;  
  LONG bottom;  
} RECT, *PRECT;""") 
 
-- replaces 'q', could be named that: 
atom pRECT = allocate_struct(idRECT) 
-- The above is still equivalent to allocate, but with size auto-calculated, 
-- and the following are still equivalent to poke(), with offset&size "". 
set_struct_field(idRECT,pRECT,"left",10) 
set_struct_field(idRECT,pRECT,"top",10) 
set_struct_field(idRECT,pRECT,"right",350) 
set_struct_field(idRECT,pRECT,"bottom",200) 

So unless anyone has any serious objections, I'm just going to delete all references to c-struct in the docs,
and clean up builtins/structs.e and pmain.e a bit, and leave it at that. You good with that?

include cffi.e 
constant idRECT = define_struct("""  

typedef struct _RECT {  
  LONG left;  
  LONG top;  
  LONG right;  
  LONG bottom;  
} RECT, *PRECT;""") 
 
-- replaces 'q', could be named that: 
atom pRECT = allocate_struct(idRECT) 
-- The above is still equivalent to allocate, but with size auto-calculated, 
-- and the following are still equivalent to poke(), with offset&size "". 
set_struct_field(idRECT,pRECT,"left",10) 
set_struct_field(idRECT,pRECT,"top",10) 
set_struct_field(idRECT,pRECT,"right",350) 
set_struct_field(idRECT,pRECT,"bottom",200) 

4. Re: Phix: c-struct

• Posted by petelomax 1 week ago
• 177 views

Sometimes I wonder why I bother, then I have a week like I've just had and I realise it's because I enjoy it!
I've been struggling with a new GUI, and finally bit the bullet at the end of November, threw 0.1 in the bin
and started a brand new 0.2: two months of hard slog with not much to show, then it suddenly burst into life.
Several days nailing down a few things and a tough time ironing out all the bugs in the new layout manager,
then this week everything I touch is just turning to gold

Me too, but a) I can't justify it, and b) what the language really deserves, and even I would use, is summat like:

#ilSTRUCT(""" 
typedef struct RECT {  
  LONG left;  
  LONG top;  
  LONG right;  
  LONG bottom;  
}""") // (so the inline assembler now understands what a RECT is, at compile-time [only]) 
#ilASM { 
  mov ecx, RECT.length 
  call :%pGetPool        // aka eax := allocate(sizeof(RECT)) 
  pragma struct RECT eax // an assembly directive, aka "eax is a *RECT" 
  mov [eax.left], 10 
  mov [eax.top], 10 
  mov [eax.right], 350 
  mov [eax.bottom], 200 
  pragma struct none eax // switch off the special eax dot handling 
} 

Which emits the perfect super-fast machine code to do these things. One day, maybe, but probably not soon.
To clarify: hll code using dot notation calling generic code in builtins/structs.e is daft compared to asm.
The #ilASM would just treat mov [eax.top], 10 as if it said mov dword[eax+4], 10 ie automate size/offset.

6. Re: Phix: c-struct

• Posted by petelomax 1 week ago
• 166 views

You got nothing to worry about there, I'm not inclined to trash anything anyone actually uses.
As you mentioned and I was only rambling on about some obscure write once forget about areas where extra effort is justifiable.