Re: Some missing features of Gtk::TreeModelColumn

[Bob Caryl <bob fis-cal com>]

Kari Pahula wrote:

> I've been going through hoops since Gtk::TreeModelColumn is missing a
> few methods, for which I see no really good reason to not be there.
>
> class FooClass
> {
> public:
>  int bar;
> };
>
> class ModelColumns : public Gtk::TreeModel::ColumnRecord
> {
> public:
>  Gtk::TreeModelColumn<int> a;
>  Gtk::TreeModelColumn<int> b;
>  Gtk::TreeModelColumn<FooClass *> c;
> };
>
> ModelColumns m_columns;
>
>  
First off, you fail to provide a constructor for your ModelColumns class 
that adds your Gtk::TreeModelColumns, ala

   ModelColumns(){add(a); add(b); add(c);};

Perhaps you omitted this for clarity;

> ...
>
> Gtk::TreeModel::Row row; // Set somewhere
> std::ifstream foo("somefile");
> foo >> row[m_columns.a];  // Fails.
>
> I have to do something like this, for no good reason:
> int tmp;
> foo >> tmp;
> row[m_columns.a] = tmp;
>
> It seems to me that with a bit of work with templates this could be
> made to work.
>
> Another problem:
> row[m_columns.a] = row[m_columns.b];	// Fails.
>
> Once again, I have to work around this.
> int tmp;
> tmp = row[m_columns.b];
> row[m_columns.a] = tmp;
>
> Is there really any good reason to have the copy
> constructor/operator=() private here?
>
> One more annoyance:
> row[m_columns.c]->bar = 5;	// Fails.
>
> Simple enough to work around with:
> (*row[m_columns.c]).bar = 5;
>
> But having operator->() in the first place would be polite.
>
> _______________________________________________
> gtkmm-list mailing list
> gtkmm-list gnome org
> http://mail.gnome.org/mailman/listinfo/gtkmm-list
>  
In the case of the std::ifstream situation, this convenience does not 
allow the __istream_type & operator>> to determine the data type in use 
by the Gtk::TreeModelColumn in question (you must supply that data along 
with the column number in a call to Gtk::TreeRow::set_value or 
Gtk::TreeRow::get_value).  Therefore, a direct assignation using the 
Gtk::TreeRow::operator[] will fail since __istream_type & operator>> 
supports only the following data types:  bool, short, unsigned short, 
int, unsigned int, long, unsigned long, float, double, long double, 
void*, and streambuf_type*.

The Gtk::TreeRow::operator[], depending on its position relative to the 
"=" operator,  merely provides a convenient syntax for invoking the 
Gtk::TreeRow::set_value and Gtk::TreeRow::get_value methods.  
Furthermore, the values supplied to this operater must match the data 
type used when initializing the Gtk::TreeModelColumn template member 
used with the operator.  These facts should amply explain why your 
attempts at assigning one operator[] call to another won't work.

Personally, I find the implementation of the Gtk::TreeRow::operator[] to 
be wonderfully convenient considering the alternative.  You would be 
much further along had the effort you expended criticizing the 
Gtk::TreeRow class implementation been applied to your own project(s) 
using the Gtk::TreeRow::operator[] as documented.

Bob Caryl