Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!wuarchive!gem.mps.ohio-state.edu!ginosko!uunet!ucscc.ucsc.edu!gorn!filbo
From: filbo@gorn.santa-cruz.ca.us (Bela Lubkin)
Newsgroups: comp.lang.pascal
Subject: Re: Clearing the keyboard buffer
Message-ID: <50.filbo@gorn.santa-cruz.ca.us>
Date: 4 Oct 89 07:30:08 GMT
References: <48.filbo@gorn.santa-cruz.ca.us> <89276.021637TBC101@PSUVM.BITNET>
Organization: R Pentomino
Lines: 40
X-Claimer: I >am< R Pentomino!

In article <89276.021637TBC101@PSUVM.BITNET> Thomas B. Collins, Jr. writes:
>In article <48.filbo@gorn.santa-cruz.ca.us>, Bela Lubkin says:
>>  While KeyPressed Do Crud:=ReadKey;  { Flush keyboard buffer }
>   If Crud = #0 then Crud := ReadKey;  { Use in case of extended key code}

>If the last character in the buffer was "extended" (Alt-something,
>etc.), Keypressed will be false, but there will still be a value in
>ReadKey to be read.  If you forget to read this, it can cause trouble
>later on...

This isn't true in any version of Turbo Pascal that I've ever used...  The
keyboard handling code has always worked as follows: call the BIOS to check
for a key.  If it's not "extended", return the value.  If it is "extended",
save the value in a temporary variable and return the prefix value (ESC before
Turbo 4.0, NUL after).  KeyPressed has always been cognizant of the saved
extended value.

In fact, in Turbo 3.0 I used code such as:

  Read(Kbd,Ch);
  Extended:=(Ch=ESC) And KeyPressed;
  If Extended Then Read(Kbd,Ch);

to read keystrokes, including extended characters.  In 4/5/5.5:

  Ch:=ReadKey;
  Extended:=(Ch=NUL) And KeyPressed;
  If Extended Then Ch:=ReadKey;

The check of KeyPressed isn't strictly necessary in 4/5/5.5 since it is very
difficult to get the IBM BIOS to generate a NUL character from the keyboard,
but is probably good practice anyway.  On the other hand, it was necessary in
Turbo 3.0 to prevent the program from blocking on the ESC key.  The 3.0 code
was susceptible to typed-ahead ESC "char" being interpreted as an extended
sequence; in programs where that really mattered, I found the appropriate
character in the runtime library and patched it to a NUL...

Bela Lubkin     * *   filbo@gorn.santa-cruz.ca.us   CIS: 73047,1112
     @        * *     ...ucbvax!ucscc!gorn!filbo    ^^^  REALLY slow [months]
R Pentomino     *     Filbo @ Pyrzqxgl (408) 476-4633 & XBBS (408) 476-4945