On Fri, Oct 26, 2001 at 09:49:42AM -0700, Terry Lambert wrote:
> Kris Kennaway wrote:
> >
> > On Thu, Oct 25, 2001 at 11:53:30PM -0700, Terry Lambert wrote:
> >
> > > I have a 10k shell script for doing this. It copies in all the
> > > stuff from the parent system which is important to match to
Kris Kennaway wrote:
>
> On Thu, Oct 25, 2001 at 11:53:30PM -0700, Terry Lambert wrote:
>
> > I have a 10k shell script for doing this. It copies in all the
> > stuff from the parent system which is important to match to the
> > real kernel, including /kernel, the "ps" program, and basically
>
Kris Kennaway wrote:
> > What I've always been wondering since Kris first mentioned this
> > technique in the thread's course (building -STABLE in a jail on
> > a -RELEASE host or vice versa, IIUC) was the following: There's
> > the host's kernel serving a differing world's userland. We all
> >
Dag-Erling Smorgrav wrote:
> The answer is that the tools used to build world *generally* aren't
> affected by changes in the kernel. The stuff that usually breaks when
> your kernel is out of synch (ps, top, ipfw...) isn't needed to build
> world. Of course, there are exceptions, like trying to
On Thu, Oct 25, 2001 at 11:53:30PM -0700, Terry Lambert wrote:
> I have a 10k shell script for doing this. It copies in all the
> stuff from the parent system which is important to match to the
> real kernel, including /kernel, the "ps" program, and basically
> all other programs that open /dev/
On Thu, Oct 25, 2001 at 10:10:11PM +0200, Gerhard Sittig wrote:
> On Wed, Oct 24, 2001 at 18:27 -0600, Warner Losh wrote:
> >
> > At timing solutions, we build all our products in a chroot jail. [...]
> >
> > We don't build RELEASES in the chroot. We build a system (make world
> > DESTDIR=xxx
Gerhard Sittig <[EMAIL PROTECTED]> writes:
> What I've always been wondering since Kris first mentioned this
> technique in the thread's course (building -STABLE in a jail on
> a -RELEASE host or vice versa, IIUC) was the following: There's
> the host's kernel serving a differing world's userland
Gerhard Sittig wrote:
>
> On Wed, Oct 24, 2001 at 18:27 -0600, Warner Losh wrote:
> >
> > At timing solutions, we build all our products in a chroot jail. [...]
> >
> > We don't build RELEASES in the chroot. We build a system (make world
> > DESTDIR=xxx outside of the chroot) that we then use t
On Wed, Oct 24, 2001 at 18:27 -0600, Warner Losh wrote:
>
> At timing solutions, we build all our products in a chroot jail. [...]
>
> We don't build RELEASES in the chroot. We build a system (make world
> DESTDIR=xxx outside of the chroot) that we then use to build the
> system (inside the ch
In message <[EMAIL PROTECTED]> Kris Kennaway writes:
: On Mon, Oct 22, 2001 at 11:58:23AM +1000, Harry Starr wrote:
: > It seems to be nigh impossible to build a "previous" release on -current.
: >=20
: > Problems include "incomplete" cross tool building, header files, and of
: > course, device su
kris> Yeah, like I said, you can do this in a jail under 5.0.
Yes, but your kernel should have 'vn' device driver which is already
deprecated in recent 5-current. Without vn, 4-stable build (make a
release) will fail when vnconfig(8) does its job for boot floppies.
Of course, we can fake vncon
On Tue, Oct 23, 2001 at 12:08:17PM +1000, Harry Starr wrote:
>
> - Original Message -
> From: "Makoto MATSUSHITA" <[EMAIL PROTECTED]>
> To: <[EMAIL PROTECTED]>
> Sent: Tuesday, October 23, 2001 10:21 AM
> Subject: Re: RELENG_4 builds on -current
&g
- Original Message -
From: "Makoto MATSUSHITA" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Tuesday, October 23, 2001 10:21 AM
Subject: Re: RELENG_4 builds on -current
> That's true, maybe he only wanna say about 4-stable "make release"
>
> Problems include "incomplete" cross tool building, header files, and
> of course, device support, in particular differences between the
> "old" vn stuff, and the new "md" device.
kris> I build worlds in a jail populated with the target release so there's
kris> no problems with this.
That's tr
On Mon, Oct 22, 2001 at 11:58:23AM +1000, Harry Starr wrote:
> It seems to be nigh impossible to build a "previous" release on -current.
>
> Problems include "incomplete" cross tool building, header files, and of
> course, device support, in particular differences
> between the "old" vn stuff, an
It seems to be nigh impossible to build a "previous" release on -current.
Problems include "incomplete" cross tool building, header files, and of
course, device support, in particular differences
between the "old" vn stuff, and the new "md" device.
Is there any intention to support cross-release
16 matches
Mail list logo
