Subject: Re: fortune makefiles incorrect ?
To: None <current-users@netbsd.org>
From: Greg A. Woods <woods@most.weird.com>
List: current-users
Date: 09/08/1999 02:25:47
[ On , September 7, 1999 at 22:25:31 (-0700), Chris G. Demetriou wrote: ]
> Subject: Re: fortune makefiles incorrect ?
>
> "Perry E. Metzger" <perry@piermont.com> writes:
> > > How many of them can't be changed (fixed?)
> > > to do something like:
> >
> > It would be unpleasant to alter every single program one wanted to
> > port with a compat library.
>
> Some would say that it's unpleasant to write inherently unportable
> code to begin with, especially if those unportabilities are hard to
> work around on other systems when porting the code. 8-)
Thank you! You took the words right from my keyboard!
Would it be so difficult to go through all the NetBSD code that uses
this fancy err() routine and add a call to properly and publicly
initialize the library code? I would think this would be a *good* thing
to do regardless of the fact that it simultaneously solves a nagging
portability problem with code in the tree! I doubt any of the
alternatives I would suggest would be anywhere near as palatable to most
*BSD'ers.
When I re-hashed newsyslog I simply gave up and stopped using err() --
there were just way far too many little gotchas to pull in without
quadrupling the size of an independent package for such a small program.
Keeping my own static "progname" and using fprintf() was infinitely
simpler even though it did indeed result in a tiny amount more typing
and infinitesimally more obscurity in the code.
--
Greg A. Woods
+1 416 218-0098 VE3TCP <gwoods@acm.org> <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>