Re: [gtkmm] Two optional return values in pango functions: what to do?

[MHL Schulze t-online de (Martin Schulze)]

Am 06.09.2002 23:14 schrieb(en) Naofumi Yasufuku:
> At Fri, 6 Sep 2002 22:39:11 +0200,
> Martin Schulze wrote:
> >
> > So far only for one pango function this has been completely implemented
> in
> > pangomm: There are three wrapper functions for
> > pango_font_get_glyph_extents:
> >
> > 1) void Font::get_glyph_extents(Glyph glyph, Rectangle& ink_rect,
> > Rectangle& logical_rect)
> > 2) void Font::get_glyph_extents_ink_rect_only (Glyph glyph, Rectangle&
> > ink_rect)
> > 3) void Font::get_glyph_extents_logical_rect_only (Glyph glyph,
> Rectangle&
> > logical_rect)
> >
> > So the return values are really optional here.
> >
> > For other funtions, like GlyphString::extents(), that also return two
> > rectangles we only have 1). Now, in my opinion, it would be an overkill
> to
> > add 2) and 3) for every such function. E.g. GlyphString::extents()
> already
> > has an overload because some parameters are optional. If we added 2)
> and 3)
> > for both overloads, we would have 6 functions in the whole that wrap
> the
> > same
> > single pango function.
> >
> > On the other hand, if people agreed that it isn't desirable to add 2)
> and
> > 3)
> > whereever possible, I would want to know whether anyone has some need
> for
> > the
> > existing functions 2) and 3) in Pango::Font. Otherwise, I would just
> remove
> > them to be consistent with the other wrapper functions that only have
> 1).
> >
>
> I think pango_font_get_glyph_extents() means that
>
>  I.   Programmers can get the ink extents of a glyph within a font.
>
>     Rectangle Font::get_glyph_ink_extents(Glyph glyph)
>
>  II.  Programmers can get the logical extents of a glyph within a font.
>
>     Rectangle Font::get_glyph_logical_extents(Glyph glyph)
>
>  III. Programmers can also get both extents at once.
>
>     void Font::get_glyph_extents(Glyph glyph,
>                                  Rectangle& ink_rect,
>                                  Rectangle& logocal_rect)
>
>
> Regards,
> --Naofumi

So you do want this convenient interface for each function that has two
return values?

Regards,

  Martin