[Qt-creator] Qt creator, gdb and anonymous unions.

[paulo]

Hi,

the py script now works for nested anonymous unions, but apparently the 
hash used in the process (inconveniently) sorts the names.
I think it would be better to keep the groupings. (just remove the sorting)
ex:

struct S {
    union {
        struct { float a, z; };
        struct { float b, y; };
    };
} s;

sould be formatted as:
$1 +
     +-+a
      |  +z
      |
     +-+b
         +x

and not
$1 +
     +a
     +b
     +y
     +z

Cheers,
Paulo

André Pönitz wrote:
 > On Sunday 14 February 2010 19:08:35 ext Andre Poenitz wrote:
 >> On Sun, Feb 14, 2010 at 09:38:26PM +0900, paulo wrote:
 >>> Hi friends at the qt-creator mailing list.
 >>>
 >>> I'm wondering if it's possible to get qt-creator debugger view to show
 >>> some anonymous unions the same way gdb does.
 >>>
 >>> That is, for something like:
 >>> struct A {
 >>>     union {
 >>>         struct { float b; };
 >>>         struct { float c; };
 >>>     };
 >>> } a;
 >>>
 >>> a.b = 0;
 >>>
 >>> gdb will print p a:
 >>> $1 = {{b = 0},{c = 0}}
 >>>
 >>> on the other hand qt-creator just says it's an anonymous union and 
will
 >>> refuse to print visualize anything, making me add it by hand to the
 >>> watchers window (very annoying).
 >> I works for a 'plain' anonymous union, but you have nested anonymous
 >> structs, too. This is an oversight on my part, I'll have a look 
tomorrow.
 >
 > Ok. This is fixed now for "python enabled" versions of gdb, both for
 > display and editing.
 >
 > For non-python gdb there's no direct solution. gdb/MI does not seem
 > to give access to the necessary data in this case. I can work around
 > and manually extract data from gdb/CLI output, but I am not sure this
 > is worth the trouble. Do you use such nested anon structures often?
 >
 > Andre'
 > _______________________________________________
 > Qt-creator mailing list
 > Qt-creator at trolltech.com
 > http://lists.trolltech.com/mailman/listinfo/qt-creator
 >