Mail Archives: djgpp/1999/10/27/13:02:55
In article <Pine DOT SUN DOT 3 DOT 91 DOT 991027160225 DOT 29100C-100000 AT is> you wrote:
> On Wed, 27 Oct 1999, Hans-Bernhard Broeker wrote:
> > This lead to a plain DJGPP compilation seeing the RSXNTDJ headers,
> > with all the troubles you get from that. The problem is that an
> > environment variable will be used by *both* RSXNTDJ and DJGPP, once
> > it's been set, and it will usually cause (at least) one of them to
> > stop working properly.
> Oh, I see that I misunderstood you. Sure, C_INCLUDE_PATH should
> include the RSXNTDJ directories *only* when you are compiling with
> RSXNTDJ.
Exactly. Which leads you to the 'setup batch file' method, either in
conjunction with separate DOS boxes startable from Windows, or by
enforcing a call to a batchfile 'setrsxnt' before you work with
RSXNTDJ, and another one to switch back to DJGPP. But with current
RSXNTDJ, quite some work seems to have gone into _avoiding_ that type
of trickery, so I think it's better not to re-introduce it for this
purpose.
As soon as there's more than one type of commandline C compiler on the
same machine, the setup batch file method becomes a necessity, of
course, but that's a different story in its own right.
> But if the user changes the value of C_INCLUDE_PATH to
> correspond to the compiler that's currently in use, then everything
> _will_ work, right?
Right. _If_.
Which is exactly the problem: users will sometimes forget to modify
their env.vars while switching back and forth between compilers, and
the symptoms of that will be hard to interpret by non-experts. Meaning
that users will come here asking about it.
Thus my rather strict advice to stay away from these environment
variables is like the usual 'Don't open this, no user serviceable
parts inside' one, Eli's one about not editing DJGPP.ENV, unless you
know very well what you're doing.
--
Hans-Bernhard Broeker (broeker AT physik DOT rwth-aachen DOT de)
Even if all the snow were burnt, ashes would remain.
- Raw text -