Re: atf for libcurses

[Rhialto <rhialto%falu.nl@localhost>]

On Fri 05 Nov 2010 at 11:02:06 +0100, Martin Husemann wrote:
> I'm with Brett here - what you propose could be done by feeding the libcurses
> output to xterm and rendering a screen shot, then comparing the bits.

Sort of, I suppose :-) But not all the world is an xterm...

> You can verify the look by this way, but nothing more. Brett's method allows
> testing of far more detailed behaviour, and is not limited to a single
> terminal emulation. This is very good.

Isn't is the case that the command sequence issued by curses depends on
the terminal's capabilities? For instance, it would use Clear To End Of
Line commands if the terminal knows them, or it would overwrite with
spaces. Direct cursor positioning commands or relative cursor movements,
depending on distance to move and if it has those commands available.
(Cursor Home + newlines + overtyping, if needed in extreme cases when
most commands are lacking).

To test all (or even a lot of) possible combinations of capabilities by
comparing command sequences is going to be A Lot Of Work(tm). And A Lot
Of Updating to do if curses' internal decisions change. Comparing the
end result (appearance) to me seems to be much simpler and more to the
point.

> Having to adapt result files after substantial changes to the inner
> workings of curses (so it produces better sequences) is no problem, and

All 2 to the power of N of them? Where N is quite large, on the order of
the number of possible capabilities in the termcap entries (each one can
be present, or not).

> Martin
-Olaf.
-- 
___ Olaf 'Rhialto' Seibert  -- There's no point being grown-up if you 
\X/ rhialto/at/xs4all.nl    -- can't be childish sometimes. -The 4th Doctor