Rainer Orth wrote:
I've recently started doing builds of OpenSolaris again, and noticed a
whole bunch of problems with SUNWonbld in the process. Stephen Lau suggested
I post a summary here before filing bug reports at bugs.opensolaris.org, so
here we go:
Wow, an amazing amount of stuff! Thanks for digging up all of this dirty
laundry, I'm sure other folks will benefit tremendously from your efforts.
Just a few things to add:
...
* The files and directories in the SUNWonbld packages distributed on
opensolaris.org belong to a non-system uid (stevenl/staff if viewed with
GNU tar), which is ugly if you extract them somewhere as root. I think
they should use cpiotranslate as well to only use system uids and gids.
Install glom kernels have the same caveat; they are best extracted with
"-o". That should probably be documented if it isn't.
* SUNWonbld should comply with Sun's own packaging guidelines and install
into /opt/SUNWonbld instead of /opt/onbld. This change requires a couple
of other places to be updated as well, but I've a list of affected files
and would obviously update them at the same time.
And everybody's .cshrc or .profile or whatever. ;)
* At one time, I had to abort nightly with ^C. Unfortunately, the
sub-dmakes ignored this an ran along happily. It took me some effort to
kill off the whole bunch ;-(
Interestingly there is nothing in the quick start guide about how to kill
off a nightly. It's actually easy because nightly now puts all of its
subprocesses into a resource container in the form of a task and the task
ID is near the top of the nightly.log file. This one-liner does it:
pkill -T `awk '/taskid/ { print $3; exit 0 }' nightly.log`
- Eric
_______________________________________________
opensolaris-code mailing list
opensolaris-code@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/opensolaris-code
