Re: Font style in widget->size_request

[David Brigada <brigad rpi edu>]

All,

I fixed my problem, and it wasn't how I had described it.  It turns out 
that I had not checked what type of units Pango was returning font sizes 
in, and I had needed to scale by the DPI of the screen.  It turns out 
that the font description *is* available after you call 
gtk_widget_ensure_style in the size_request method.

Sorry about the confusion,
David Brigada

Paul Davis wrote:
> On Tue, Mar 24, 2009 at 2:18 PM, David Brigada <brigad rpi edu 
> <mailto:brigad rpi edu>> wrote:
>
>     Hello,
>
>     I am writing a graph widget for use in GTK+ applications.  I'm
>     trying to calculate the minimum size for the widget from the text
>     along the axes of the graph.  I use Cairo to draw the text, so I'm
>     using the cairo_text_extents functions to calculate their sizes.
>
>     However, I'm running into a problem where widget->style->font_desc
>     does not seem to be returning a valid Pango font description.  When
>     I call pango_font_description_get_size on the font_desc, it returns
>     10 * PANGO_SCALE, which seems to be a default value when the style
>     isn't initialized.  This causes my function to assume the incorrect
>     font size when calculating its size request.  I tried calling
>     gtk_widget_ensure_style on my widget before accessing the font_desc
>     element, but that does not help.
>
>
> at some point the development of GTK (i don't recall precisely where, 
> but roughly in the vicinity of GTK 2.10), a change was made to the way 
> styles "bind" to widgets. the key change was that a widget that is not 
> attached to a GtkWindow doesn't have a style (yet).
>
> i have several examples in my code where i need to do style-based size 
> computation, and so I create a dummy GtkWindow, pack some (temporary, 
> representative) widget into it (I do not show the window), then use 
> ensure_style, and then do the computation, and then delete them both. 
> the critical change was adding the widget to the (dummy) window. it was 
> initially a very irritating change, but in the end a pretty trivial code 
> change.
>
> ---p