Subject: Re: checkflist error codes
To: Frederick Bruckman <>
From: Greywolf <>
List: current-users
Date: 11/05/2003 10:34:12
Thus spake Frederick Bruckman ("FB> ") sometime Today...

FB> On Wed, 5 Nov 2003, Greywolf wrote:
FB> > Example:  I want to build a distribution with no /lib, /libexec or
FB> > /rescue, and a statically-linked root. Or I might want to build a pure
FB> > static distribution.  Or I might want to build a distribution that uses
FB> > compressed manual pages.  Or I might want to build a distribution which
FB> > will install nicely onto an embedded installation.
FB> >
FB> > But checkflist bites my knee when I do that.  I find myself adding '-V
FB> > MAKEFLAGS=-i' to the build process, something which, as even a
FB> > lightweight-by-comparison developer/programmer, I *really* don't like
FB> > doing.
FB> The build could be a little smarter, and set ${NOPOSTINSTALL} if
FB> various MK* options are set to "no". For yourself, can't you simply

Hi, Frederick,

Certainly no affront meant, but this is like trying to place a band-aid
over a shotgun wound.  There should be several flist files conditionally
tossed into the final mix, and checkflist should run against that, rather
than running against the Whole Big Default Picture.

For the moment, I'm sure ${NOPOSTINSTALL} would work (I wasn't even aware
of that variable, even after having gone through extensive work to make
mk.conf do The Right Things WRT the rest of the other flags (SHAREDSTRINGS
is still broken -- from closer examination, it looks as though the COPTS
or CFLAGS or something which it needs is getting ignored in that case)),
but something more fine-grained needs to exist.

If I have the time to delve into it before someone else does, I will do
so, but if someone else has the time and inclination, that would work,
too. :)

NetBSD: Twice the Bits-Clean of other Leading OSes.