Created attachment 148039 [details]
This is the code showing the bug

This is the code showing the bug

Also reported at http://linuxtesting.org/results/report?num=D0142

Comment on attachment 148039 [details]
This is the code showing the bug

changing the mime type

I think the documentation is quite clear about that:

"if the default size does not obey the geometry hints for the window (gtk_window_set_geometry_hints() can be used to set these explicitly), the default size will be clamped to the nearest permitted size."

And even more exactly

"Windows can't actually be 0x0 in size, they must be at least 1x1, but passing 0 for width and height is OK, resulting in a 1x1 default size."

(In reply to comment #4)
> I think the documentation is quite clear about that:
> 
> "if the default size does not obey the geometry hints for the window
> (gtk_window_set_geometry_hints() can be used to set these explicitly), the
> default size will be clamped to the nearest permitted size."
> 
> And even more exactly
> 
> "Windows can't actually be 0x0 in size, they must be at least 1x1, but passing
> 0 for width and height is OK, resulting in a 1x1 default size."

So it would be better to change the documentation and don't confuse developers with 0 value, because as you mentioned windows can't be 0 sized and 0 has no any special meaning.

(In reply to comment #4)
> I think the documentation is quite clear about that:
> 
> "if the default size does not obey the geometry hints for the window
> (gtk_window_set_geometry_hints() can be used to set these explicitly), the
> default size will be clamped to the nearest permitted size."
> 
> And even more exactly
> 
> "Windows can't actually be 0x0 in size, they must be at least 1x1, but passing
> 0 for width and height is OK, resulting in a 1x1 default size."

So it would be better to change the documentation and don't confuse developers with 0 value, because as you mentioned windows can't be 0 sized and 0 has no any special meaning.