[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Multi-OS Makefiles

On Sat, Aug 07, 2004 at 01:08:06PM -0400, Ted Unangst wrote:
> > > Still running my good old home-brewn linux box, I know that auto-hell is
> > > fine *if* it works.
> > 
> > How often have you seen it fail compare to work ?
> > Do you get 1 failure in 100, 1 in 10 or 1 in 1000 ?
> trying to compile software without using patches from ports?  at least 1 
> in 10.  if libtool is involved, my experience goes up closer to 1 in 1.

It's interesting that for *me* the failure count is still between 2 and
5 in 100. That's much less than your estimation, but remember that I'm
talking about a linux system, not a BSD. It seems as if lots of auto-stuff
is written (or copied & pasted), used and tested on well-configured
linux boxes, i.e. "standard" distributions, only.

So, if a software package builds on *some* linux systems only, why at
all use auto-hell?

> as compared with a bogus LIBS+=-ldl i frequently find in hand crafted 
> makefiles (the only fix necessary for the program to work), fixing 
> configure when it doesn't work is a nightmare.  unless the author has done 
> it right (and they rarely do), [...]

That's the point. Lots of people are using autohell, but only few of
them really understand how it really works (I don't, either). It's too
complicated, there are too many incompatible releases (I've *four*
versions of autoconf on my new OpenBSD system after building and
installing a hand full of ports).

Are those developers too dumb to properly use autohell? I don't think so.
It's the tools that causes problems, not the developers.


Visit your host, monkey.org