Newsgroups: comp.lang.pascal
Path: utzoo!utgpu!watserv1!maytag!watstat.waterloo.edu!dmurdoch
From: dmurdoch@watstat.waterloo.edu (Duncan Murdoch)
Subject: Re: TP Units
Message-ID: <1991Jun11.002648.14082@maytag.waterloo.edu>
Sender: news@maytag.waterloo.edu (News Owner)
Organization: University of Waterloo
References: <1991Jun9.150801.5178@minyos.xx.rmit.oz.au> <60Bc41w164w@ersys.edmonton.ab.ca>
Date: Tue, 11 Jun 1991 00:26:48 GMT
Lines: 33

In article <60Bc41w164w@ersys.edmonton.ab.ca> bns@ersys.edmonton.ab.ca (Chris Dollmont) writes:
>
>I'm I the only one that is upset by the fact that Borland keeps changing 
>the unit structure so I have to upgrade EVERYTHING, not just my TP??

I think you've got it backwards.  Borland keeps changing the unit structure,
because all of your .TPU files would be useless anyways.  Every time a change
is made to the interface of a unit, anything that uses it has to be 
recompiled.  I can't imagine a new version of TP that wouldn't have changes
to the SYSTEM unit, let alone the other standard units.  

For example:  TP6 and TPW 1.0 have identical .TPU formats, but of course
a .TPU file from one is useless in the other.

You're right to complain about this, but you can protect yourself
against it in two ways.  The way I do it is to never, ever, rely on a unit
unless I have the source code.  (Actually, external routines in an .OBJ
file are reasonably safe.)  If I have the source code, I'm confident that
I'll be able to recompile it in a new version (though I may need to make
some changes).  If I don't, I rely on the good will and success at remaining
solvent of the supplier of the unit.

The other way to protect yourself is just not to upgrade TP.  If a lot of
people chose this option, Borland might get the message.

By the way, it now looks possible to get cut off in the same way in C as
well as TP.  I'd bet that the precompiled header files that Borland's
recently introduced there are as version specific as .TPU files; we'll see
if anyone starts distributing .OBJ+precompiled headers now that it's 
possible.

Duncan Murdoch
dmurdoch@watstat.waterloo.edu