Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ames!lll-winken!uunet!tektronix!percival!qiclab!m2xenix!dawggon!p101.f19.n490.z2.fidonet.org!Pat.Terry
From: Pat.Terry@p101.f19.n490.z2.fidonet.org (Pat Terry)
Newsgroups: comp.lang.modula2
Subject: -32768
Message-ID: <1242.2459235C@dawggon.fidonet.org>
Date: 29 Apr 89 05:02:49 GMT
Sender: ufgate@dawggon.fidonet.org (newsout1.26)
Organization: FidoNet node 2:490/19.101 - Settler City Fido, Grahamstown RSA
Lines: 19

There has been considerable discussion here recently about
      I := -32768;
      WriteInt(I,10)

This is handled "correctly" by JPI, QuickMod and FTL compilers.

FST objects to I := -32768.

I put the point to Roger Carvalho, the FST author, and his comment was
that -32768 might validly be regarded by a compiler as an expression rather  
than as a literal.  Now -32768 = -(32768), and 32768 is an invalid INTEGER  
literal on the machines we are considering, so he has a point.  This does  
not seem to have been mentioned in the discussion yet.
  


--  
uucp: {mcvax!uunet,tektronix,sun!nosun}!oresoft!dawggon!2!490!19.101!Pat.Terry
Internet: Pat.Terry@p101.f19.n490.z2.fidonet.org