>
> Reasoning:
>
> The main documentation would stay where most of the code is. In the
> future this might change slightly because we will need to split the
> device drawing code in different files for different platforms, but
> IMHO still most of the drawing code and the docs could stay in the
> base Fl_Device class. Deviations and additions would then have to
> be documented in the derived classes (Fl_GDI_Display etc.).
>
> We should make the mechanics of drawing by different Fl_*_Device
> classes public anyway, so that there is a clear interface how to
> add other devices (e.g. frame buffer).
>
> Drawback: it could be confusing to have to switch from fl_draw()
> to Fl_Device::draw() in the docs to find something, but once you
> know how it is designed, this would probably be "normal".
>
> Note: this is only "thinking loud", I don't know if it is feasible.
> Please comment.
>
> Albrecht

I've committed a series of changes to fully document the Fl_Device,
Fl_Abstract_Printer and Fl_Printer classes.

This is done putting comments in .H files. I'm conscious that this
bends the rule, but this rule is not really practical for virtual
classes where a member function has several implementations,
so there is not a single place where the code for this function is.
At present, there are two codes for each drawing function: one
for the 'classical' display output, one for PostScript output.
In the future, there'll be, hopefully, still others.
I also noted that several instances of similar rule bending were
present in the documentation (e.g., line 218 of FL/fl_draw.H)
for the very same reason: avoiding duplication.

If the result is not satisfactory, I'm ready to consider other
solutions.
_______________________________________________
fltk-dev mailing list
[email protected]
http://lists.easysw.com/mailman/listinfo/fltk-dev

Reply via email to