Re: Gtk::VariantIter replacement

[Murray Cumming <murrayc murrayc com>]

On Fri, 2011-04-01 at 23:54 +0300, Povietkin Konstantin wrote:
> В Птн, 01/04/2011 в 22:26 +0200, Murray Cumming пишет:
> > On Fri, 2011-04-01 at 21:15 +0300, Povietkin Konstantin wrote:
> > > Gtk::VariantIter is useful only for classes, which are derived from
> > > Gtk::VariantContainerBase class. Has not it more sense to make
> > > Gtk::VariantIter more STL-styled?
> > > 
> > > For example:
> > > Gtk::VariantContainerBase::iterator iter;
> > > Gtk::VariantContainerBase::const_iterator const_iter;
> > > 
> > > In my opinion current usage of Gtk::VariantIter is useful only for
> > > non-OOP languages.
> > > Am I wrong?
> > 
> > A patch would be welcome, but we must freeze the API very soon. For now,
> > it's probably just a simple wrapper around the C API.
> > 
> I can try to write a patch, but I have not enough experience to do this.
> So I think it'll be better if it do somebody more experienced.

The variantiter.hg file has this comment, maybe from me:
"
//This can't be like a real iterator (like Gtk::TextIter),
//because g_iter_value_get_next_value() both gets a value and changes
the iterator.
//GtkTextIter allows us to go forward and then separately get the
current value.
"
See
http://library.gnome.org/devel/glib/unstable/glib-GVariant.html#g-variant-iter-next-value

I guess it's possible, but it would probably be inefficient, involving
extra copies of the VariantIter. But there's no time for that now, and
it would probably be best to improve the underlying GVariantIter,
hopefully by just making API additions. Feel free to open a bug about
it.

-- 
murrayc murrayc com
www.murrayc.com
www.openismus.com