Re: [gtk-list] Re: [patch] GtkNotebook

[Tim Janik <Tim Janik Hamburg Netsurf DE>]

On Sun, 1 Feb 1998, Lars Hamann wrote:

> 
> On Sat, Jan 31, 1998 at 01:46:20AM +0100, Tim Janik wrote:
> 
> > this is a problem with other widgets as well, from the TODO:
> >  * Display of GtkToggleButton is messed up if GtkContainer::container_width
> >    is greate than 0. GtkCheckButton and GtkRadioButton are only messed up
> >    if draw_indicator is FALSE.
> 
> OK, fixed. ;)

thanx, haven't looked at it yet, though.

> > > [FIX] (gtk_notebook_page_allocate) :
> > >    the tab label is sized too big (by CHILD_SPACING)
> > >    e.g. GTK_POS_TOP : 
> > >       tab_label->allocation.height was
> > >       tab_label->requisition.height + CHILD_SPACING
> > 
> > shouldn't that be a parameter that can be specified as a widget argument?
> 
> I don't know. Would be easy to implement, but I'm not sure if someone
> needs it.

it would be nice if that could be set from outside the toolit.

> > > [FIX] (gtk_notebook_remove_page) :
> > >     call gtk_notebook_switch_page if "cur_page" is removed
> > 
> > this should be done in gtk_notebook_insert_page also,
> > actually the signal needs to be emitted in all places where cur_page is
> > assigned a new value i.e.
> > gtk_notebook_insert_page,
> > gtk_notebook_remove_page and
> > gtk_notebook_current_page.
> 
> OK, gtk_notebook_insert_page should "switch_page", i'll fix this.
> But i can't see, why gtk_notebook_current_page should do that ?

i actually take the "switch_page" signal as "current_page_notify".
maybe the "switch_page" name is a bit missleading, but it should be
invoked as soon as the current page changes (i.e. notebook->cur_page
is altered).

> > also the switch_page signal should be extended to also pass the page number,
> > because the notebooks interface i.e.
> > void       gtk_notebook_remove_page     (GtkNotebook      *notebook,
> >                                          gint              page_num);
> > void       gtk_notebook_set_page        (GtkNotebook      *notebook,
> >                                          gint              page_num);
> > expects page numbers and not page pointers:
> >   void (* switch_page)       (GtkNotebook *notebook,
> >                               GtkNotebookPage *page,
> >                             gint             page_num);
> 
> Sounds good.
>  
> > > Some suggestions :
> > > 
> > > - How about a popup menu on the tab section for quick page switching ?
> > >   This could be done by an additional "gchar *" parameter to 
> > >   gtk_notebook_append_page, insert_page, prepend_page. This string will
> > >   appear in the popup, describing the contents of the page.
> > 
> > why couldn't this string just be the notebook label?
> 
> Because GtkWidget *tab_label is more powerful. It's easy to use
> a box with a Pixmap and Text or something completely different. Maybe
> it would even make sense to add another GtkWidget instead of a "gchar *".

would be nice to have, but there is quite some code out there, that uses
the current interfae to gtknotebook.
what about:
void       gtk_notebook_prepend_page_with_item (GtkNotebook      *notebook,
	                                        GtkWidget        *child,
        	                                GtkWidget        *tab_label,
						GtkWIdget	 *menu_item);

or somesuch. the ordinary gtk_notebook_prepend_page will then just create
a menu item with a label that duplicates tab_label->label.
we would provide the interface compatibility this way and still have your
enhancements.

> > > - As the properties "scrollable", "show_tabs", "show_border" are 
> > >   separate styles, how about combining gtk_notebook_set_show_tabs,
> > >   set_show_border, set_scrollable into one function gtk_notebook_set_style,
> > >   taking an enum as argument ?
> > 
> > a "widget-style" is already something else, look at gtkwidget.[hc] and
> > gtkstyle.[hc] to see what *_set_style is used for.
>  
> OK, let's say ... gtk_notebook_set_appearance ... or so.
> The Question with both suggestions is, if it's OK to break existing code.

well that depends. gtk+ has broken existing code for quite a few times,
but never without proper reasoning. i personally consider your note_book
enhancements not important enough to adjust all existing applications
for them. this does not imply i would find your enhancements nonsense or
unusable. but as long as the interface compatibility can be provided,
we should do.
no more concret, gtk_notebook_set_appearance is certainly a good idea,
but it doesn't rule out the old functions that accesed each item solely,
am i wrong? ;)

> > > - Should gtk_notebook_foreach be expanded to also iterate over the
> > >   tab label widgets ? These are certainly also children of the
> > >   container...
> > 
> > hm, this issue is a little bit tricky.
> > gtkwidget.c relies on certain behaviours on the container_foreach
> > functions, and i'm not sure gtknotebook fits that either way...
> 
> Could you explain this a little bit ?
> I can't see why tab_label should not be processed by 
> gtk_widget_set_style_recurse for instance.

yea, you are probably right, it'll just require proper care with the new
refcounting stuff. if you are going to provide a patch for this, please
wait until the next gtk version is out, or work from a recent cvs snap shot.

> 
> bye,
>   Lars
> 

---
ciaoTJ