Re: [ros-dev] [ros-diffs] [akhaldi] 65249: [SHELL32] * bool => BOOL.

Sun, 09 Nov 2014 21:36:59 -0800 


On 2014-11-10 05.19, Timo Kreuzer wrote:
The point is: It is different. A C++ bool will always be only 1 or 0. MSVC does actually warn (a performance warning) when assigning an int to a bool, GCC does not warn. And it's ok, since it will (should) do the right thing. 

I'm glad GCC have the good sense to not complain ;)
That's what I wrote the bool_cast function to deal with in MSVC...
so you don't have to litter your code with warning disabling pragmas.


A bool will always be 0 or 1. A BOOL could be anything. So if there is 
code relying on a parameter to be 0 or 1, since it was a bool, it 
might fail, when it gets a BOOL, which might be 0x80000000.
"if (FooBar == TRUE)" might fail, if FooBar is converted from bool to 
BOOL.


I agree..

I strongly favor bool, which is better than BOOL for a few reasons I can 
think of.



1) It's *safe*, unlike BOOL, even for explicit truth-value compares 
(which are truly bad habit).

The following will work with bool, but fail with BOOL:

volatile DWORD Flags = 0xFFFFFFFF;

bool  ok = bool_cast( Flags & 0x80000000 );
if (ok == TRUE) puts( "Passed" ); else puts( "Failed" ); // Passed

BOOL  B52 = Flags & 0x80000000;
if (B52 == TRUE) puts( "Passed" ); else puts( "Failed" ); // Failed


2) The compiler can (potentially) generate more efficient binary code 
for bool.



It might not be an issue. And I don't say we should not use BOOL or 
BOOLEAN, I just want to point out that these are not the same thing 
and you cannot just convert one to the other and expect that 
everything works as before.


It *is* a potential issue, but perhaps an unexpected one..

We *must forbid* explicit truth-value comparisons, since they are 
bombers, per the above.
(The "logical not" operator (!) exists for a reason.. I'm sure we can 
all agree on that. ;)



On 2014-11-10 07.56, David Quintana (gigaherz) wrote:

All the common COM interfaces use BOOL instead of bool, because it's 
more compiler-agnostic and it matches better the Win32 API type usage. 
If we use bool, we may risk a "consumer" of the interface using the 
wrong size for "bool", so although ISFHelper is an internal interface 
not meant to be used by 3rdparty programs, BOOL just fits better. It 
wasn't wrong before, it was just "out of place" kinda.





The compiler will always push a 32 bit boolean value, regardless if its 
bool or BOOL,

so it should not be a concern as far as the size is concerned.

More importantly, as Timo originally stated, and I illuminated above,
changing bool to BOOL can potentially introduce code malfunctions,
which I'm sure none of us would like.


Best Regards
// Love




Am 09.11.2014 17:28, schrieb Love Nystrom:

@Timo, @Amine


A BOOL should merely distinguish between zero and non-zero, so 
extending an 8 bit bool
(which I favor) to a 32 bit BOOL should always produce a valid 
result, even in the example case.
If Foo becomes true after the mask test, it is expressed as non-zero, 
which persist through the
BOOLEAN, and still remain non-zero when expanded to BOOL, so either 
should be OK,

provided you don't need to comply with someone elses interface.

As arguments, the compiler will always pass 32 bit in either case 
(afaik).


But MSVC *will* whine about assigning FlagsValue & 0x80000000 to a bool.

I made a bool_cast function (below) to deal with those cases, since 
programmers
usually know what they do when they choose to squeeze a BOOL into a 
bool.

true == TRUE, which we know, but the compiler whines and bitches about.
I'm not sure how to deal with gcc for this, but it ought be similar.


Masking 0x80000000 directly to BOOLEAN will fail of course, which the 
compiler

detects: conversion from 'DWORD' to 'BOOLEAN', possible loss of data

#ifdef _MSC_VER
// This cast is forced inline, and should not generate any extra code,
// just coax the compiler to treat the BOOL as a bool without whining.

#pragma warning( disable: 4800 )
bool __forceinline bool_cast( BOOL value ) {
    return (bool) value;
}
#pragma warning( default: 4800 )
#endif

Then you are free to do:


bool Foo = bool_cast( Flags & 0x80000000 ); // Nice, quiet, and 
compact ;-)


Use it carefully though.

Just my penny to the bucket..
Best Regards
// Love

On 2014-11-05 05.06, Timo Kreuzer wrote:



There is a difference between a C++ bool and a win32 BOOL. bool is 8 
bit and aware of things like (bool Foo = FlagsValue & 0x80000000); 
BOOLEAN BooleanFlag = Foo;" with BOOL this would fail. I don't 
really know whether a BOOL is really the right thing to use here.


Am 04.11.2014 21:00, schrieb akha...@svn.reactos.org:

Author: akhaldi
Date: Tue Nov  4 20:00:09 2014
New Revision: 65249

URL: http://svn.reactos.org/svn/reactos?rev=65249&view=rev
Log:
[SHELL32]
* bool => BOOL.

Modified:

branches/shell-experiments/dll/win32/shell32/folders/CDesktopFolder.cpp 


branches/shell-experiments/dll/win32/shell32/folders/CDesktopFolder.h
branches/shell-experiments/dll/win32/shell32/folders/CFSFolder.cpp
branches/shell-experiments/dll/win32/shell32/folders/CFSFolder.h
branches/shell-experiments/dll/win32/shell32/shellfolder.h


Modified: 
branches/shell-experiments/dll/win32/shell32/folders/CDesktopFolder.cpp 

URL: 
http://svn.reactos.org/svn/reactos/branches/shell-experiments/dll/win32/shell32/folders/CDesktopFolder.cpp?rev=65249&r1=65248&r2=65249&view=diff
============================================================================== 

--- 
branches/shell-experiments/dll/win32/shell32/folders/CDesktopFolder.cpp 
[iso-8859-1] (original)
+++ 
branches/shell-experiments/dll/win32/shell32/folders/CDesktopFolder.cpp 
[iso-8859-1] Tue Nov  4 20:00:09 2014

@@ -1189,7 +1189,7 @@
      return ret;
  }

  -HRESULT WINAPI CDesktopFolder::CopyItems(IShellFolder *pSFFrom, 
UINT cidl, LPCITEMIDLIST *apidl, bool bCopy)
+HRESULT WINAPI CDesktopFolder::CopyItems(IShellFolder *pSFFrom, 
UINT cidl, LPCITEMIDLIST *apidl, BOOL bCopy)

  {
      CComPtr<IPersistFolder2> ppf2;
      WCHAR szSrcPath[MAX_PATH];


Modified: 
branches/shell-experiments/dll/win32/shell32/folders/CDesktopFolder.h
URL: 
http://svn.reactos.org/svn/reactos/branches/shell-experiments/dll/win32/shell32/folders/CDesktopFolder.h?rev=65249&r1=65248&r2=65249&view=diff
============================================================================== 

--- 
branches/shell-experiments/dll/win32/shell32/folders/CDesktopFolder.h 
[iso-8859-1] (original)
+++ 
branches/shell-experiments/dll/win32/shell32/folders/CDesktopFolder.h 
[iso-8859-1] Tue Nov  4 20:00:09 2014

@@ -88,7 +88,7 @@

          virtual HRESULT WINAPI GetUniqueName(LPWSTR pwszName, 
UINT uLen);
          virtual HRESULT WINAPI AddFolder(HWND hwnd, LPCWSTR 
pwszName, LPITEMIDLIST *ppidlOut);
          virtual HRESULT WINAPI DeleteItems(UINT cidl, 
LPCITEMIDLIST *apidl);
-        virtual HRESULT WINAPI CopyItems(IShellFolder *pSFFrom, 
UINT cidl, LPCITEMIDLIST *apidl, bool bCopy);
+        virtual HRESULT WINAPI CopyItems(IShellFolder *pSFFrom, 
UINT cidl, LPCITEMIDLIST *apidl, BOOL bCopy);

            DECLARE_REGISTRY_RESOURCEID(IDR_SHELLDESKTOP)
          DECLARE_NOT_AGGREGATABLE(CDesktopFolder)


Modified: 
branches/shell-experiments/dll/win32/shell32/folders/CFSFolder.cpp
URL: 
http://svn.reactos.org/svn/reactos/branches/shell-experiments/dll/win32/shell32/folders/CFSFolder.cpp?rev=65249&r1=65248&r2=65249&view=diff
============================================================================== 

--- 
branches/shell-experiments/dll/win32/shell32/folders/CFSFolder.cpp 
[iso-8859-1] (original)
+++ 
branches/shell-experiments/dll/win32/shell32/folders/CFSFolder.cpp 
[iso-8859-1] Tue Nov  4 20:00:09 2014

@@ -1030,7 +1030,7 @@
   * copies items to this folder
   */

  HRESULT WINAPI CFSFolder::CopyItems(IShellFolder * pSFFrom, UINT 
cidl,
-                                    LPCITEMIDLIST * apidl, bool 
bCopy)
+                                    LPCITEMIDLIST * apidl, BOOL 
bCopy)

  {
      CComPtr<IPersistFolder2> ppf2 = NULL;
      WCHAR szSrcPath[MAX_PATH];


Modified: 
branches/shell-experiments/dll/win32/shell32/folders/CFSFolder.h
URL: 
http://svn.reactos.org/svn/reactos/branches/shell-experiments/dll/win32/shell32/folders/CFSFolder.h?rev=65249&r1=65248&r2=65249&view=diff
============================================================================== 

--- 
branches/shell-experiments/dll/win32/shell32/folders/CFSFolder.h 
[iso-8859-1] (original)
+++ 
branches/shell-experiments/dll/win32/shell32/folders/CFSFolder.h 
[iso-8859-1] Tue Nov  4 20:00:09 2014

@@ -102,7 +102,7 @@

          virtual HRESULT WINAPI GetUniqueName(LPWSTR pwszName, 
UINT uLen);
          virtual HRESULT WINAPI AddFolder(HWND hwnd, LPCWSTR 
pwszName, LPITEMIDLIST *ppidlOut);
          virtual HRESULT WINAPI DeleteItems(UINT cidl, 
LPCITEMIDLIST *apidl);
-        virtual HRESULT WINAPI CopyItems(IShellFolder *pSFFrom, 
UINT cidl, LPCITEMIDLIST *apidl, bool bCopy);
+        virtual HRESULT WINAPI CopyItems(IShellFolder *pSFFrom, 
UINT cidl, LPCITEMIDLIST *apidl, BOOL bCopy);

            DECLARE_REGISTRY_RESOURCEID(IDR_SHELLFSFOLDER)
          DECLARE_NOT_AGGREGATABLE(CFSFolder)

Modified: branches/shell-experiments/dll/win32/shell32/shellfolder.h

URL: 
http://svn.reactos.org/svn/reactos/branches/shell-experiments/dll/win32/shell32/shellfolder.h?rev=65249&r1=65248&r2=65249&view=diff
============================================================================== 

--- branches/shell-experiments/dll/win32/shell32/shellfolder.h 
[iso-8859-1] (original)
+++ branches/shell-experiments/dll/win32/shell32/shellfolder.h 
[iso-8859-1] Tue Nov  4 20:00:09 2014

@@ -42,7 +42,7 @@
      STDMETHOD(GetUniqueName)(THIS_ LPWSTR  lpName, UINT uLen) PURE;

      STDMETHOD(AddFolder)(THIS_ HWND  hwnd, LPCWSTR lpName, 
LPITEMIDLIST * ppidlOut) PURE;
      STDMETHOD(DeleteItems)(THIS_ UINT  cidl, LPCITEMIDLIST * 
apidl) PURE;
-    STDMETHOD(CopyItems)(THIS_ IShellFolder * pSFFrom, UINT cidl, 
LPCITEMIDLIST * apidl, bool bCopy) PURE;
+    STDMETHOD(CopyItems)(THIS_ IShellFolder * pSFFrom, UINT cidl, 
LPCITEMIDLIST * apidl, BOOL bCopy) PURE;

  };
  #undef INTERFACE










_______________________________________________
Ros-dev mailing list
Ros-dev@reactos.org
http://www.reactos.org/mailman/listinfo/ros-dev








_______________________________________________
Ros-dev mailing list
Ros-dev@reactos.org
http://www.reactos.org/mailman/listinfo/ros-dev






















					Reply via email to