On 20.05.2012 13:30, Armin Le Grand wrote:
Hi Pavel,
Pavel Janík<pa...@janik.cz> wrote:
Hi,
WaE = Warning as Error.
or Warnings Are Errors.
Gcc option -Werror: Make all warnings into errors.
With this option turned on, all warnings are made into errors. Our long
term goal is to make gcc silent.
Ah, yes, I remember now. We spent some time on this years ago, but never
coud make the whole code WaE-safe.
I thought that most of non-binfilter code was WaE-safe.
Anyway, we should still try to keep/make our code WaE-safe. It can only
improve its quality.
Despite that, it's good to work on it;
sometimes this gives good hints at weird code.
These issues are not errors per se, but e.g.:
@@ -1330,6 +1331,7 @@
void ImpSdrGDIMetaFileImport::DoAction(MetaWallpaperAction& rAct)
Please just comment /*rAct*/
{
+ (void) rAct;
OSL_ENSURE(false, "Tried to construct SdrObject from
MetaWallpaperAction: not supported (!)");
}
This means that rAct is unused in the method. gcc warns about it.
This change:
@@ -1384,6 +1388,7 @@
case GRADIENT_ELLIPTICAL: aXGradientStyle = XGRAD_ELLIPTICAL;
break;
case GRADIENT_SQUARE: aXGradientStyle = XGRAD_SQUARE; break;
case GRADIENT_RECT: aXGradientStyle = XGRAD_RECT; break;
+ default: break;
Hmm. I do not have the code at hand right now, cannot tell until monday.
}
const XFillGradientItem aXFillGradientItem(
means that some enum value is forgotten in the switch.
This change:
- for(sal_uInt32 y(0); y< pOld->Height(); y++)
+ for(sal_Int32 y(0); y< pOld->Height(); y++)
Please change to sal_uInt32
Why? Does not look like y could become negative (assuming that
pOld->Height() is also a sal_uInt32.
means that we were comparing signed and unsigned value.
I do not know if these changes are OK, thus I send the patch as I used to
make the module WaE free.
I have not seen a patch. If you have one, please send again (maybe
directly) and I'll happily take a look on monday. I'm currently not
compiling on gcc, so I will not be able to guarantee, though.
Hope this helps.
P.S. Of course warnings differ between compilers and sometims the changes look
weird etc.
Yes, I remember now. Do you have a good solution for swich..case where not
all missing entries would have to be listed?
Pavel has already done that above (add "default: break;" to your switch
statement.
Who will check in the changes?
-Andre
