Bug 25769 – Fix accessibility warnings on GTK

• Status: RESOLVED
• Resolution: FIXED
• Component: Accessibility

[chris fleizach]

here are the warnings this patch will try to fix

../../WebCore/page/AccessibilityTable.h:66: warning: type qualifiers ignored on function return type
../../WebCore/page/AccessibilityTable.h:67: warning: type qualifiers ignored on function return type
../../WebCore/page/AccessibilityRenderObject.cpp:2118: warning: ‘WebCore::createARIARoleMap()::RoleEntry’ declared with greater visibility than the type of its field ‘WebCore::createARIARoleMap()::RoleEntry::ariaRole’
../../WebCore/page/AccessibilityTable.h:66: warning: type qualifiers ignored on function return type
../../WebCore/page/AccessibilityTable.h:67: warning: type qualifiers ignored on function return type
../../WebCore/page/AccessibilityTable.cpp:353: warning: type qualifiers ignored on function return type
../../WebCore/page/AccessibilityTable.cpp:361: warning: type qualifiers ignored on function return type
../../WebCore/page/AccessibilityTable.h:66: warning: type qualifiers ignored on function return type
../../WebCore/page/AccessibilityTable.h:67: warning: type qualifiers ignored on function return type
../../WebCore/page/AccessibilityTable.h:66: warning: type qualifiers ignored on function return type
../../WebCore/page/AccessibilityTable.h:67: warning: type qualifiers ignored on function return type

[chris fleizach]

Created attachment 30301 [details]
patch

[Mark Rowe (bdash)]

Comment on attachment 30301 [details]
patch

This isn't correct.  The warnings are saying that it doesn't make sense for the return type of a function to be "const unsigned", such as in "const unsigned columnCount();".  Should that perhaps be "unsigned columnCount() const;"?

[chris fleizach]

yes that sounds correct

[chris fleizach]

Created attachment 30350 [details]
patch

updated patch based on review
can't make columnCount into

unsigned columnCount() const

because it needs to call addChildren(), which is not const

[Holger Freyther]

Comment on attachment 30350 [details]
patch


> Index: WebCore/page/AccessibilityRenderObject.cpp
> ===================================================================
> --- WebCore/page/AccessibilityRenderObject.cpp	(revision 43716)
> +++ WebCore/page/AccessibilityRenderObject.cpp	(working copy)
> @@ -2116,6 +2116,7 @@ typedef HashMap<String, AccessibilityRol
>  static const ARIARoleMap& createARIARoleMap()
>  {
>      struct RoleEntry {
> +    public:


Isn't a struct public by default?

[chris fleizach]

you'd think, but what does this warning mean then

‘WebCore::createARIARoleMap()::RoleEntry’ declared with greater visibility
than the type of its field

[Mark Rowe (bdash)]

*** Bug 25838 has been marked as a duplicate of this bug. ***

[Darin Adler]

Comment on attachment 30350 [details]
patch

>  {
>      struct RoleEntry {
> +    public:
>          String ariaRole;
>          AccessibilityRole webcoreRole;
>      };

This change does not make sense. In a struct, everything is public by default and this should have no effect. I have no idea why gcc is warning about this, but I do not think this is the fix.

> Index: WebCore/page/AccessibilityTable.cpp
> ===================================================================
> --- WebCore/page/AccessibilityTable.cpp	(revision 43716)
> +++ WebCore/page/AccessibilityTable.cpp	(working copy)
> @@ -350,7 +350,7 @@ void AccessibilityTable::cells(Accessibi
>      }
>  }
>      
> -const unsigned AccessibilityTable::columnCount()
> +unsigned AccessibilityTable::columnCount()
>  {
>      if (!hasChildren())
>          addChildren();
> @@ -358,7 +358,7 @@ const unsigned AccessibilityTable::colum
>      return m_columns.size();    
>  }
>      
> -const unsigned AccessibilityTable::rowCount()
> +unsigned AccessibilityTable::rowCount()
>  {
>      if (!hasChildren())
>          addChildren();
> Index: WebCore/page/AccessibilityTable.h
> ===================================================================
> --- WebCore/page/AccessibilityTable.h	(revision 43716)
> +++ WebCore/page/AccessibilityTable.h	(working copy)
> @@ -63,8 +63,8 @@ public:
>      AccessibilityChildrenVector& columns();
>      AccessibilityChildrenVector& rows();
>      
> -    const unsigned columnCount();
> -    const unsigned rowCount();
> +    unsigned columnCount();
> +    unsigned rowCount();
>      
>      virtual String title() const;

These changes are good.

I'm saying r=me, but don't land that "public:" change -- it should have no effect.

[chris fleizach]

alright, i won't make the struct change, which makes no sense (as does the warning). someone who compiles gtk can fix that

[Darin Adler]

(In reply to comment #9)
> alright, i won't make the struct change, which makes no sense (as does the
> warning).

I think that warning indicates a bug in the new gcc. Someone should investigate more and report the bug to the gcc maintainers.

[Darin Adler]

http://trac.webkit.org/changeset/43810