[Libreoffice] [PATCH] Removed dependencies on tools/solar.h in drawpage.hxx/.cxx

[Norbert Thiebaud]

On Wed, Jan 5, 2011 at 12:56 PM, Soeren Moeller
<soerenmoeller2001 at gmail.com> wrote:
> Hi and thank you for your comments
>
> 2011/1/5 Kohei Yoshida <kyoshida at novell.com>:
>> Hi Sören,
>>>
>>> +#include <sal/types.h>
>>>  #include "address.hxx"
>>
>> This should not be necessary (though it won't hurt) as the address.hxx
>> already includes sal types.  Does it not build without this change for
>> you?
> It does build without the change for me, I did the change to ensure
> that we don't break the file accidentially by changing includes of
> address.hxx, although I now can see that this maybe isn't necessary
> for such basic includes (types are quite standard and almost everyone
> includes address.hxx). I think I felt it more important yesterday
> after reading a lot of files using things from tools/ but only
> including them through other includes, which would break the code, if
> I would remove the tools/ parts from the other includes first.

I think that including what you need where you need it is a good practice.
replying on unrelated included to bring 'magically' the other includes
you need is a maintenance pain.
(with the exception of a set of known and documented 'meta-include',
which we do not really have)

I do think that solar.h related stuff should be such a mandatory
meta-include that should be the first include of every sources of
ours.
That would make easy to deal with type migration and
compatibility-hack, by giving an entry point where to put these.

Norbert

>
>>
>>>  // Let's put here misc structures that get passed to ScDocument's methods.
>>> diff --git a/sc/inc/drawpage.hxx b/sc/inc/drawpage.hxx
>>> index faa43fa..5e207b7 100644
>>> --- a/sc/inc/drawpage.hxx
>>> +++ b/sc/inc/drawpage.hxx
>>> @@ -30,7 +30,7 @@
>>>  #define SC_DRAWPAGE_HXX
>>>
>>>  #include <svx/fmpage.hxx>
>>> -
>>> +#include <sal/types.h>
>>>
>>>  class ScDrawLayer;
>>>
>>> @@ -39,7 +39,7 @@ class ScDrawLayer;
>>>  class ScDrawPage: public FmFormPage
>>>  {
>>>  public:
>>> -    ScDrawPage(ScDrawLayer& rNewModel, StarBASIC* pBasic, BOOL bMasterPage=FALSE);
>>> +    ScDrawPage(ScDrawLayer& rNewModel, StarBASIC* pBasic, sal_Bool bMasterPage=sal_False);
>>>      ~ScDrawPage();
>>>
>>>      virtual ::com::sun::star::uno::Reference< ::com::sun::star::uno::XInterface > createUnoPage();
>>> diff --git a/sc/source/core/data/drawpage.cxx b/sc/source/core/data/drawpage.cxx
>>> index d489d5d..5a91540 100644
>>> --- a/sc/source/core/data/drawpage.cxx
>>> +++ b/sc/source/core/data/drawpage.cxx
>>> @@ -44,7 +44,7 @@
>>>
>>>  // -----------------------------------------------------------------------
>>>
>>> -ScDrawPage::ScDrawPage(ScDrawLayer& rNewModel, StarBASIC* pBasic, BOOL bMasterPage) :
>>> +ScDrawPage::ScDrawPage(ScDrawLayer& rNewModel, StarBASIC* pBasic, sal_Bool bMasterPage) :
>>>      FmFormPage(rNewModel, pBasic, bMasterPage)
>>>  {
>>>      SetSize( Size( LONG_MAX, LONG_MAX ) );
>>
>> For these two, let's use the standard bool type.  In fact, its parent
>> class FmFormPage does expect a parameter of the bool type, not sal_Bool
>> nor BOOL (though the default value is for some reason set to sal_False
>> instead of false, which should really be fixed...), which provides
>> another reason to use bool, not sal_Bool.
> I think I'm still too afraid of breaking UNO-things, if there just is
> something UNO-related close to the code, when using native types, I
> will try to be more confident in using bool, and believing in it, if
> my build works out fine.
>
> Regards
> Sören
>
>
>>
>> Thanks!
>>
>> Kohei
>>
>> --
>> Kohei Yoshida, LibreOffice hacker, Calc
>> <kyoshida at novell.com>
>>
>>
> _______________________________________________
> LibreOffice mailing list
> LibreOffice at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/libreoffice
>