Just to be clear, Yes, I think it is extremely poor engineering to try to 
evolve build environments and the systems that they build without an 
explicit mechanism for a) identifying the dependency and b) bootstapping it 
during a build cycle. I would not take issue with this crud if, like it has 
been done on other versions of *BSD, 'make build' would detect that the 
build tools being used would not build the release, then build the required 
tools into a "safe" place (like $DESTDIR/build_env), then fix up the paths 
in the makefile, and _then_ build the system using the new tools.

And while I realize this was a fairly mild rebuke to my complaint, I stress 
that I managed to both complain and identify the issue and its fix in just 
three sentences. Not so with this one, sorry.

Finally, I have addressed the "powers that be", such as they are, and they 
don't care. They aren't doing this "professionally" and they either can't 
appreciate, or they don't understand the benefit of a reliable build 
system. I would normally attribute this to inexperience but I know at least 
some of them have more build hours in their past than I do. Perhaps they 
haven't experienced the power of what a truely automated build can give 
you. Imagine every single supported architecture building every night with 
every configuration and regression tested when the build finished! There 
are currently sufficient monkeys, such as myself, that poke at this 
manually to achieve the same thing but it pains me none the less.

Except for the occasional rant (such as this one) I pretty much ignore it. 
Please bear with the finger pointing. It disallows the defense "But this 
hasn't caused any problems really." and keeps a record in the archive.

--Chuck


At 05:41 PM 6/10/00 -0400, Entropy wrote:
> >Date: Sat, 10 Jun 2000 02:38:29 -0700
> >From: Chuck McManis <cmcmanis@mcmanis.com>
> >
> >Just did a cvs update and tried to build. Lots of bombs! Seems make has
> >been hacked to do some more magik and the build was changed to depend on
> >it. Sigh, nobody ever learns.
>
>Perhaps because you keep complaining about this on port-vax.  I can't
>say I agree with your philosophy that the build tools should never
>change before a release cycle has been completed, but you might get
>more satisfaction if you discussed your concerns in a forum where it
>might be noticed by people with the ability to change those
>practices.  I would think that current-users and/or tech-userlevel
>might be more appropriate.
>
>--
>entropy -- it's not just a good idea, it's the second law.