(making patch)

Created attachment 45945 [details] [review]
(proposed patch)

(Ok this time I actually patched the relevant .hg and .ccg files rather than
the gmmproc output, sorry for my ignorance of the build process.)

This is confusing because create*() methods are usually static, instead of using
an instance. Thinking about it, it also does not seem much more convenient
anyway. I'm tempted to reject this.

You're right, a non-static function shouldn't be called create*, would including
it as e.g. subpixbuf(...) be a better option though? Because if you think nabout
it, everywhere else in gtkmm C functions that operate on an instance are
generally wrapped as member functions, with constructors and create methods
wrapping C functions that construct something -- this much is obvious and makes
sense -- but gdk_pixbuf_new_subpixbuf seems to fit both categories, it accepts
an instance as its first argument /and/ constructs another.

To be honest I found it odd that the feature is a static create method at all,
which imho feels unnatural to use. It doesn't feel like constructing, and if
gtkmm didn't have need of create methods, I doubt it would've been implemented
as a constructor.

Also, saying that
  pixbuf->subpixbuf(...)                      (object->method(args))
isn't much more convenient than
  Gdk::Pixbuf::create_subpixbuf(pixbuf, ...)  (objecttype_method(object, args))
seems like saying that e.g.
  button->set_relief(Gtk::RELIEF_NONE)
isn't much more convenient than
  gtk_button_set_relief(button, GTK_RELIEF_NONE)

Anyway, thanks for your time. Sorry to be bothering you with such trivial things.

*It doesn't feel like constructing any more than scale_simple() does (sorry, I
switched windows to check scale_simple wasn't also a static function, then
forgot to type it here)

Well, it can't be called create_*(). Do you have another suggestion?

Please reopen if you can think of anything appropriate. If not, it doesn't seem essential.