Re: vcl checkbox behaviour

On 28/06/12 12:44, Jan Holesovsky wrote:
> Hi Noel,
>
> Noel Power píše v Po 25. 06. 2012 v 11:46 +0100:
>
> > I am somewhat stumbling over what would be best to do, change the
> > underlying vcl class unconditionally ? ( that would imply the uno
> > controls behaviour also changes unconditionally ) or ... if we change
> > the underlying vcl class conditionally what about the uno controls?,
> > should we change them also conditionally? or unconditionally? It would
> > seem to me if there is some consensus that the 'normal' vcl behaviour is
> > not that normal but is 'abnormal' then I would be inclined to change the
> > vcl class unconditionally ( and similarly the behaviour of the dependant
> > uno controls ) Any thoughts ?
> So the behavior sound curious to me; but OTOH nearly sounds as if
> somebody did that deliberately.  Do you have pointers to the code where
> does the "apply the Align property on the control itself" live?  Any
> history that would be of some worth around there?
no like I mentioned earlier in the mail the behaviour is indeed 
intentional see 
http://opengrok.libreoffice.org/xref/core/vcl/source/control/button.cxx#3209 
However, I just don't understand why it, historically this behaviour 
seems to be there for quite a while, the last change around there was
http://cgit.freedesktop.org/libreoffice/core/commit/?id=d727e1094312e34fedc2bfdb12b43c109a954054 
which adds a new logic leg  for checkboxes with images ( which is a 
strange concept for me too :-) )