DragonFly commits List (threaded) for 2005-03
Re: cvs commit: src/share Makefile src/share/doc Makefile src/usr.sbin/lpr Makefile
On Tue, 15 Mar 2005 00:18:06 +0100
User Joerg <joerg@xxxxxxxxxxxxxxxxx> wrote:
> On Mon, Mar 14, 2005 at 09:03:52AM -0800, Chris Pressey wrote:
> > I'm OK with this (I actually like the idea of pregenerating stuff in
> > CVS in general.)
> I'm not a fan of commiting pregenereated stuff into the CVS repo.
> Actually I want to continue removing prebuild stuff like the
> various device tables and create them on the fly.
Yeah, I understand. I was thinking of two things in particular:
- having pre-built .depend files would shave off some buildworld time,
but yes, it would be more of a hassle when committing stuff, to make
sure the .depend files are up to date.
(Alternatively, maybe make(1) itself could be trained to build the
.depend files itself when they don't exist...? Any thoughts, Max?)
- the monster known as bin/sh. It would be generous to describe it as
being written in C; it's written in some ad-hoc programming language
which is translated into C by its build process :) Again, maybe this
could be improved upon by some means besides just committing a
pregenerated version into the repo, but, as it stands, it is one very
intimidating piece of software, and I don't really like that, since it
encourages a sort of "don't touch it, it might break" attitude for
> > I'm also OK with an (e.g.) "installextradocs" target, to be run
> > after "installworld", which generates and installs these docs.
> That's something I don't object at all :) Fell free to add such a
> target to etc/Makefile. I think that's the location making the most
> sense. Well, actually it should be "buildextradocs" and
> "installextradocs" :)
Right. I may or may not get around to this any time soon, since I don't
feel terribly strongly about it, but at least it's an open option.