Re: gboolean definition and size

[Dan Nicolaescu <dann godzilla ICS UCI EDU>]

Havoc Pennington <hp redhat com> writes:

  > Dan Nicolaescu <dann godzilla ICS UCI EDU> writes:
  > > Havoc Pennington <hp redhat com> writes:
  > > 
  > >   > Dan Nicolaescu <dann godzilla ICS UCI EDU> writes: 
  > >   > > Is there any reason for gboolean to have a different size? 
  > >   > 
  > >   > IIRC it used to be a byte, but there were dozens of bugs introduced by
  > >   > that, so we just gave up and made it an int.
  > > 
  > > So is it a good time now to revisit that decision? 
  > > Or you don't think it's appropriate? 
  > > 
  > 
  > I think we'll be to GTK 3 by the time C99 is widely deployed enough to
  > be interesting, most likely.

OK.

  > I almost always use "guint foo : 1" bitfields for booleans anyhow,
  > when space is an issue...

Then please replace the following gboolean with a bitfield in:

struct _GtkCellRenderer
{
  GtkObject parent;

  gboolean can_activate;
  gboolean visible;

[snip]
}

Maybe the GtkCellRenderer could use the same trick as GtkWidget for
storing these flags in "parent"... That would save 8 bytes. I could
send a patch for that if desired...


        --dan