DragonFlyBSD

:> mkdep -f .depend -a =20
:...
:> /usr/obj/usr/src/world_i386/usr/include/time.h:48:29: =20
:> machine/uvparam.h: No such file or directory
:> mkdep: compile failed
:> *** Error code 1
:
:Is that fixed?
:
:cheers
: simon

I'll do a buildworld and nrelease run and fix whatever problems crop
    up.

Is your snap build doing a buildworld followed by a buildkernel?  Or
    is just doing a 'nativekernel' build?  You may have to manually update
    the 'make' and 'config' programs, or build and install a new world.

-Matt

But it's buildworld that is failing. The `make' program inherits ${MACHINE}
from either the environment variable with the same name, or from
`sysctl -n hw.machine'. So if you try to run buildworld of HEAD on a
pre-infrastructure-change kernel (either 1.6.x-RELEASE or 1.7.x-DEVELOPMENT),
you'll get this error because header files in /sys/machine/pc32/include
won't get installed into ${WORLDDEST}, because ${MACHINE} expands to i386,
not pc32 as expected. Doing something like

env MACHINE=pc32 make buildworld

got me past this stage, but wouldn't we like to avoid having to do this
in the next release?

:But it's buildworld that is failing. The `make' program inherits ${MACHINE}
:from either the environment variable with the same name, or from
:`sysctl -n hw.machine'. So if you try to run buildworld of HEAD on a
:pre-infrastructure-change kernel (either 1.6.x-RELEASE or 1.7.x-DEVELOPMENT),
:you'll get this error because header files in /sys/machine/pc32/include
:won't get installed into ${WORLDDEST}, because ${MACHINE} expands to i386,
:not pc32 as expected. Doing something like
:
: env MACHINE=pc32 make buildworld
:
:got me past this stage, but wouldn't we like to avoid having to do this
:in the next release?

I think the solution here is to simply detect the situation and have
    the buildworld abort with a message telling the user what they have
    to do.  I'll do it right now.

-Matt
                    Matthew Dillon 
                    <dillon@backplane.com>

Wouldn't it be better to warn the user, sleep for a couple of seconds,
then set it to pc32, then continue the build?

Cheers,
--
Thomas E. Spanjaard
tgen@netphreax.net

:Wouldn't it be better to warn the user, sleep for a couple of seconds,
:then set it to pc32, then continue the build?
:
:Cheers,
:--
: Thomas E. Spanjaard
: tgen@netphreax.net

No.  It's a one-time deal.  All that needs to be done is to not leave
    the user scratching his head wondering what to do.  We don't want to
    try to outsmart him... we want the build to stop cold so the problem
    can be corrected.

-Matt

Working again:
http://chlamydia.fs.ei.tum.de/pub/DragonFly/snapshots/i386/ISO-IMAGES/2CSNAP-20061118-2330-Devel.iso.bz2