Subject: Re: CVS commit: src/usr.bin/make
To: r.o.s.s <ross@NetBSD.org>
From: James Chacon <jmc@NetBSD.org>
List: source-changes
Date: 05/08/2004 02:51:26
On Fri, May 07, 2004 at 08:55:28PM -0000, r.o.s.s wrote:
> >
> > >Log Message:
> > >Simplify build, no functional changes.
> >
> > Since you were touching just about every file, wouldn't it have been
> > better to perhaps re-org the includes and have make.h or something
> > define a simple __RCSID() ?
>
> It's not quite that simple, the define also controls cdefs.h and
> although it could have been structured to move the stupid __RCSID()
> down below make.h that would also have involved touching every file,
> and multiple edits where there are repeated tests of the define.
>
> IMHO, cdefs.h and __RCSID() are refinements for native code whose
> negatives greatly outweigh the vague positives when interpreted in
> the context of the very first required bootstrapping tool.
>
> What I would really like to do is remove that fluff entirely from
> make(1), but I limited myself to just redoing the original MAKE_BOOTSTRAP
> edits with improvements.
>
> > BTW, I'm about to commit a change for posix compliance that I wanted
> > to get pulled up to 2.0, is this commit going to interfere with
> > that (pullup)? Ie. does this commit now also need to pulled up to 2.0?
>
> No, and no. You may or may not be able to directly apply your patch
> to 2.0 but I'm sure any changes to get it to cleanly apply to 2.0 sources
> will be trivial. Remember, pullups aren't supposed to say "pull up everything"
> or "sync with current" but identify specific delta's or provide patch(1) input.
Yes, but with intervening changes (especially ones that change code layout)
often times "pullup rev X" doesn't work directly and requires a patch.
Avoiding that if possible is always nice (though in this case pulling this
change up also wouldn't be all that bothersome)
James