Please describe the problem:
Hi there,

I know it's unavoidable in projects like Evolution, yet it's always nice to go
over code and fix the occurrences of unavoidable mistakes.

One such mistake (well, it's not really a mistake but makes replacing the
GtkHtml widget a lot more difficult) is violation of data hiding.

For example:


  <mystruct.h>

  typedef struct {
	int hidden_a;
  } MyStruct;

  int mystruct_get_a (MyStruct *instance);


  <myprogram.c>

  #include <mystruct.h>

  int test (MyStruct *struc)
  {
	printf ("It has %d\n", struc->hidden_a);
  }

Of course should the programmer here have used this:

int test (MyStruct *struc) 
{
	printf ("It has %d\n", mystruct_get_a (struc));
}

What can be done about this (I know most Evolution developers know this, but I'm
repeating it to illustrate what I did in this patch):

  <mystruct.h>

  typedef struct _MyStructPrivate MyStructPrivate;
  typedef struct {
	MyStructPrivate *priv;
  } MyStruct;

  int mystruct_get_a (MyStruct *instance);

  <mystruct.c>
  struct _MyStructPrivate {
    int hidden_a;
  };

  int mystruct_get_a (MyStruct *instance)
  {
     return instance->priv->hidden_a;
  }


That is basically what I did with e-msg-composer.c and e-msg-composer.h


The reason why I also like this to get accepted (and tested) is because this
will be the basis of replacing the GtkHtml widget with a GtkMozEmbed widget.

You see, if I'm sure that NOBODY used anything else but the agreed EMsgComposer
API as defined in e-msg-composer.h, then I can securely replace the internals of
e-msg-composer.c.

I can then for example replace the Bonobo/CORBA stuff with normal method calls
on a GObject/GtkWidget. And then also replace the GtkHtml widget with a
GtkMozEmbed that has been set editable (check our go-evolution wiki for more
information about that).

Steps to reproduce:


Actual results:


Expected results:


Does this happen every time?


Other information: