Re: Keyboard input time out routine

["Matthew Heaney" <matthew_heaney@acm.org>]

In article 
<B94C2AACE5FD49E5.F586390D9ABF88C3.A2C5492FFC18140C@lp.airnews.net> ,
Ray <crwhite12@airmail7.net>  wrote:

> Thanks Matthew, I was able to expand your information to get a working idea of
> what to do.  Looks like I need to write a procedure (like I did in C++ using
> kbhit and getch) to process any keystrokes and build the number, and time out
> if the enter key is not hit in the allotted time.

That's right.  Predefined I/O packages are "potentially blocking," so
you don't really know whether they'll abort right away.  For example,
Get could be implemented like this:

  procedure Get (Item : out Num) is
    pragma Abort_Deferred;  -- GNAT-specific pragma
  begin
    ...
  end Get;

in which case Get won't abort, even if the delay expires.

In your example, Get didn't abort.  I coded your example using Get_Line,
and it did abort.  But you can't depend on this.

The best thing to do is as you suggest: assemble the lexeme yourself by
consuming one character at a time using Get_Immediate, and terminate
when you consume the line terminator (or too much time passes).

You don't need ATC at all (for this problem).

--
Science is, foremost, a method of interrogating reality: proposing
hypotheses that seem true and then testing them -- trying, almost
perversely, to negate them, elevating only the handful that survive to
the status of a theory. Creationism is a doctrine, whose adherents are
interested only in seeking out data that support it.

George Johnson, NY Times, 15 Aug 1999